{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 8.897296094358518,
  "eval_steps": 100,
  "global_step": 5000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "entropy": 0.7850796665996314,
      "epoch": 0.017803493935684877,
      "grad_norm": 4.125,
      "learning_rate": 1.8e-07,
      "loss": 0.9868,
      "mean_token_accuracy": 0.7752813071012497,
      "num_tokens": 259048.0,
      "step": 10
    },
    {
      "entropy": 0.7656301472336053,
      "epoch": 0.035606987871369754,
      "grad_norm": 3.984375,
      "learning_rate": 3.8e-07,
      "loss": 0.9514,
      "mean_token_accuracy": 0.778770961612463,
      "num_tokens": 532019.0,
      "step": 20
    },
    {
      "entropy": 0.7800006279721856,
      "epoch": 0.053410481807054634,
      "grad_norm": 3.546875,
      "learning_rate": 5.800000000000001e-07,
      "loss": 0.9645,
      "mean_token_accuracy": 0.777244720607996,
      "num_tokens": 807220.0,
      "step": 30
    },
    {
      "entropy": 0.8148440960794687,
      "epoch": 0.07121397574273951,
      "grad_norm": 3.578125,
      "learning_rate": 7.8e-07,
      "loss": 0.9845,
      "mean_token_accuracy": 0.7725338555872441,
      "num_tokens": 1068429.0,
      "step": 40
    },
    {
      "entropy": 0.7980734214186669,
      "epoch": 0.08901746967842439,
      "grad_norm": 3.328125,
      "learning_rate": 9.800000000000001e-07,
      "loss": 0.9502,
      "mean_token_accuracy": 0.7774997223168612,
      "num_tokens": 1336698.0,
      "step": 50
    },
    {
      "entropy": 0.827414046600461,
      "epoch": 0.10682096361410927,
      "grad_norm": 2.65625,
      "learning_rate": 1.1800000000000001e-06,
      "loss": 0.969,
      "mean_token_accuracy": 0.7773337867110968,
      "num_tokens": 1608574.0,
      "step": 60
    },
    {
      "entropy": 0.8639046084135771,
      "epoch": 0.12462445754979415,
      "grad_norm": 2.21875,
      "learning_rate": 1.3800000000000001e-06,
      "loss": 0.9913,
      "mean_token_accuracy": 0.7705458387732506,
      "num_tokens": 1872747.0,
      "step": 70
    },
    {
      "entropy": 0.8748010536655784,
      "epoch": 0.14242795148547902,
      "grad_norm": 2.234375,
      "learning_rate": 1.5800000000000001e-06,
      "loss": 0.9868,
      "mean_token_accuracy": 0.7721451628953219,
      "num_tokens": 2133173.0,
      "step": 80
    },
    {
      "entropy": 0.8339586811140179,
      "epoch": 0.1602314454211639,
      "grad_norm": 2.546875,
      "learning_rate": 1.7800000000000001e-06,
      "loss": 0.9473,
      "mean_token_accuracy": 0.782527657970786,
      "num_tokens": 2406145.0,
      "step": 90
    },
    {
      "entropy": 0.8617484034970403,
      "epoch": 0.17803493935684878,
      "grad_norm": 2.671875,
      "learning_rate": 1.98e-06,
      "loss": 0.9705,
      "mean_token_accuracy": 0.7755408465862275,
      "num_tokens": 2663709.0,
      "step": 100
    },
    {
      "epoch": 0.17803493935684878,
      "eval_chemistry_entropy": 0.8530604543685913,
      "eval_chemistry_loss": 0.9879733920097351,
      "eval_chemistry_mean_token_accuracy": 0.7785031309127808,
      "eval_chemistry_num_tokens": 2663709.0,
      "eval_chemistry_runtime": 34.6174,
      "eval_chemistry_samples_per_second": 14.444,
      "eval_chemistry_steps_per_second": 3.611,
      "step": 100
    },
    {
      "epoch": 0.17803493935684878,
      "eval_physics_entropy": 0.6429356033802033,
      "eval_physics_loss": 0.7288391590118408,
      "eval_physics_mean_token_accuracy": 0.8237355589866638,
      "eval_physics_num_tokens": 2663709.0,
      "eval_physics_runtime": 41.2328,
      "eval_physics_samples_per_second": 12.126,
      "eval_physics_steps_per_second": 3.032,
      "step": 100
    },
    {
      "entropy": 0.8313142383471132,
      "epoch": 0.19583843329253367,
      "grad_norm": 2.3125,
      "learning_rate": 2.1800000000000003e-06,
      "loss": 0.931,
      "mean_token_accuracy": 0.7826357394456863,
      "num_tokens": 2934980.0,
      "step": 110
    },
    {
      "entropy": 0.8383694937452674,
      "epoch": 0.21364192722821854,
      "grad_norm": 2.140625,
      "learning_rate": 2.38e-06,
      "loss": 0.9256,
      "mean_token_accuracy": 0.7815470688045025,
      "num_tokens": 3208702.0,
      "step": 120
    },
    {
      "entropy": 0.8249719947576523,
      "epoch": 0.2314454211639034,
      "grad_norm": 2.3125,
      "learning_rate": 2.5800000000000003e-06,
      "loss": 0.9319,
      "mean_token_accuracy": 0.7833116251975298,
      "num_tokens": 3480119.0,
      "step": 130
    },
    {
      "entropy": 0.8376072259619832,
      "epoch": 0.2492489150995883,
      "grad_norm": 2.171875,
      "learning_rate": 2.7800000000000005e-06,
      "loss": 0.9339,
      "mean_token_accuracy": 0.7807847507297992,
      "num_tokens": 3746392.0,
      "step": 140
    },
    {
      "entropy": 0.8562109092250466,
      "epoch": 0.26705240903527316,
      "grad_norm": 2.125,
      "learning_rate": 2.9800000000000003e-06,
      "loss": 0.945,
      "mean_token_accuracy": 0.7766421258449554,
      "num_tokens": 4019487.0,
      "step": 150
    },
    {
      "entropy": 0.8114860905334353,
      "epoch": 0.28485590297095803,
      "grad_norm": 2.3125,
      "learning_rate": 3.1800000000000005e-06,
      "loss": 0.9108,
      "mean_token_accuracy": 0.7848536405712366,
      "num_tokens": 4298100.0,
      "step": 160
    },
    {
      "entropy": 0.8167605808004736,
      "epoch": 0.30265939690664295,
      "grad_norm": 1.984375,
      "learning_rate": 3.3800000000000007e-06,
      "loss": 0.896,
      "mean_token_accuracy": 0.7858319569379091,
      "num_tokens": 4568116.0,
      "step": 170
    },
    {
      "entropy": 0.8315373951569199,
      "epoch": 0.3204628908423278,
      "grad_norm": 2.359375,
      "learning_rate": 3.58e-06,
      "loss": 0.9106,
      "mean_token_accuracy": 0.7813973911106586,
      "num_tokens": 4831334.0,
      "step": 180
    },
    {
      "entropy": 0.8178198751062155,
      "epoch": 0.3382663847780127,
      "grad_norm": 2.015625,
      "learning_rate": 3.7800000000000002e-06,
      "loss": 0.881,
      "mean_token_accuracy": 0.7869020272046328,
      "num_tokens": 5103301.0,
      "step": 190
    },
    {
      "entropy": 0.8403754932805896,
      "epoch": 0.35606987871369755,
      "grad_norm": 1.9921875,
      "learning_rate": 3.980000000000001e-06,
      "loss": 0.9115,
      "mean_token_accuracy": 0.7783690184354782,
      "num_tokens": 5362915.0,
      "step": 200
    },
    {
      "epoch": 0.35606987871369755,
      "eval_chemistry_entropy": 0.8292876372337341,
      "eval_chemistry_loss": 0.9164837598800659,
      "eval_chemistry_mean_token_accuracy": 0.7834693474769592,
      "eval_chemistry_num_tokens": 5362915.0,
      "eval_chemistry_runtime": 34.57,
      "eval_chemistry_samples_per_second": 14.463,
      "eval_chemistry_steps_per_second": 3.616,
      "step": 200
    },
    {
      "epoch": 0.35606987871369755,
      "eval_physics_entropy": 0.6445197880268096,
      "eval_physics_loss": 0.6974144577980042,
      "eval_physics_mean_token_accuracy": 0.8239417452812194,
      "eval_physics_num_tokens": 5362915.0,
      "eval_physics_runtime": 41.2388,
      "eval_physics_samples_per_second": 12.124,
      "eval_physics_steps_per_second": 3.031,
      "step": 200
    },
    {
      "entropy": 0.8109689498320222,
      "epoch": 0.3738733726493824,
      "grad_norm": 2.078125,
      "learning_rate": 4.18e-06,
      "loss": 0.856,
      "mean_token_accuracy": 0.788726344704628,
      "num_tokens": 5625455.0,
      "step": 210
    },
    {
      "entropy": 0.8134721567854285,
      "epoch": 0.39167686658506734,
      "grad_norm": 2.0,
      "learning_rate": 4.38e-06,
      "loss": 0.8568,
      "mean_token_accuracy": 0.7867508839815855,
      "num_tokens": 5890806.0,
      "step": 220
    },
    {
      "entropy": 0.8066105756908655,
      "epoch": 0.4094803605207522,
      "grad_norm": 2.203125,
      "learning_rate": 4.58e-06,
      "loss": 0.8275,
      "mean_token_accuracy": 0.7900486279278993,
      "num_tokens": 6158751.0,
      "step": 230
    },
    {
      "entropy": 0.8255353083834052,
      "epoch": 0.4272838544564371,
      "grad_norm": 2.125,
      "learning_rate": 4.78e-06,
      "loss": 0.8244,
      "mean_token_accuracy": 0.7881908807903528,
      "num_tokens": 6423747.0,
      "step": 240
    },
    {
      "entropy": 0.8402099519968033,
      "epoch": 0.44508734839212194,
      "grad_norm": 1.9609375,
      "learning_rate": 4.980000000000001e-06,
      "loss": 0.8386,
      "mean_token_accuracy": 0.7834979113191366,
      "num_tokens": 6684517.0,
      "step": 250
    },
    {
      "entropy": 0.8010248441249133,
      "epoch": 0.4628908423278068,
      "grad_norm": 2.0,
      "learning_rate": 5.18e-06,
      "loss": 0.7988,
      "mean_token_accuracy": 0.7915425561368465,
      "num_tokens": 6945659.0,
      "step": 260
    },
    {
      "entropy": 0.814286389760673,
      "epoch": 0.48069433626349173,
      "grad_norm": 2.28125,
      "learning_rate": 5.380000000000001e-06,
      "loss": 0.8029,
      "mean_token_accuracy": 0.7904389422386885,
      "num_tokens": 7217797.0,
      "step": 270
    },
    {
      "entropy": 0.785382822342217,
      "epoch": 0.4984978301991766,
      "grad_norm": 2.0,
      "learning_rate": 5.580000000000001e-06,
      "loss": 0.76,
      "mean_token_accuracy": 0.7953414976596832,
      "num_tokens": 7498182.0,
      "step": 280
    },
    {
      "entropy": 0.8195194514468312,
      "epoch": 0.5163013241348615,
      "grad_norm": 1.90625,
      "learning_rate": 5.78e-06,
      "loss": 0.7916,
      "mean_token_accuracy": 0.78877494931221,
      "num_tokens": 7761686.0,
      "step": 290
    },
    {
      "entropy": 0.8369470547884703,
      "epoch": 0.5341048180705463,
      "grad_norm": 2.375,
      "learning_rate": 5.98e-06,
      "loss": 0.8066,
      "mean_token_accuracy": 0.786370538175106,
      "num_tokens": 8026923.0,
      "step": 300
    },
    {
      "epoch": 0.5341048180705463,
      "eval_chemistry_entropy": 0.818053927898407,
      "eval_chemistry_loss": 0.8182278871536255,
      "eval_chemistry_mean_token_accuracy": 0.7884931750297547,
      "eval_chemistry_num_tokens": 8026923.0,
      "eval_chemistry_runtime": 34.5992,
      "eval_chemistry_samples_per_second": 14.451,
      "eval_chemistry_steps_per_second": 3.613,
      "step": 300
    },
    {
      "epoch": 0.5341048180705463,
      "eval_physics_entropy": 0.6585752744674682,
      "eval_physics_loss": 0.6424084305763245,
      "eval_physics_mean_token_accuracy": 0.8238574590682983,
      "eval_physics_num_tokens": 8026923.0,
      "eval_physics_runtime": 41.2472,
      "eval_physics_samples_per_second": 12.122,
      "eval_physics_steps_per_second": 3.031,
      "step": 300
    },
    {
      "entropy": 0.8153910035267472,
      "epoch": 0.5519083120062312,
      "grad_norm": 2.6875,
      "learning_rate": 6.18e-06,
      "loss": 0.792,
      "mean_token_accuracy": 0.7901755437254906,
      "num_tokens": 8301506.0,
      "step": 310
    },
    {
      "entropy": 0.8126581776887178,
      "epoch": 0.5697118059419161,
      "grad_norm": 1.96875,
      "learning_rate": 6.380000000000001e-06,
      "loss": 0.7938,
      "mean_token_accuracy": 0.7923727743327618,
      "num_tokens": 8559126.0,
      "step": 320
    },
    {
      "entropy": 0.8185926586389541,
      "epoch": 0.587515299877601,
      "grad_norm": 2.375,
      "learning_rate": 6.5800000000000005e-06,
      "loss": 0.7873,
      "mean_token_accuracy": 0.7911712639033794,
      "num_tokens": 8821088.0,
      "step": 330
    },
    {
      "entropy": 0.7809942720457912,
      "epoch": 0.6053187938132859,
      "grad_norm": 1.8046875,
      "learning_rate": 6.780000000000001e-06,
      "loss": 0.75,
      "mean_token_accuracy": 0.7993631161749363,
      "num_tokens": 9089866.0,
      "step": 340
    },
    {
      "entropy": 0.790563733316958,
      "epoch": 0.6231222877489707,
      "grad_norm": 2.03125,
      "learning_rate": 6.98e-06,
      "loss": 0.7666,
      "mean_token_accuracy": 0.79404122158885,
      "num_tokens": 9358866.0,
      "step": 350
    },
    {
      "entropy": 0.7880068564787507,
      "epoch": 0.6409257816846556,
      "grad_norm": 1.8203125,
      "learning_rate": 7.180000000000001e-06,
      "loss": 0.7467,
      "mean_token_accuracy": 0.7980206806212664,
      "num_tokens": 9629977.0,
      "step": 360
    },
    {
      "entropy": 0.7765095194801688,
      "epoch": 0.6587292756203404,
      "grad_norm": 1.921875,
      "learning_rate": 7.3800000000000005e-06,
      "loss": 0.7403,
      "mean_token_accuracy": 0.799276452511549,
      "num_tokens": 9897757.0,
      "step": 370
    },
    {
      "entropy": 0.7657832968980074,
      "epoch": 0.6765327695560254,
      "grad_norm": 1.96875,
      "learning_rate": 7.58e-06,
      "loss": 0.7391,
      "mean_token_accuracy": 0.8021788071841002,
      "num_tokens": 10168367.0,
      "step": 380
    },
    {
      "entropy": 0.7575604457408189,
      "epoch": 0.6943362634917103,
      "grad_norm": 2.03125,
      "learning_rate": 7.78e-06,
      "loss": 0.7274,
      "mean_token_accuracy": 0.803612269833684,
      "num_tokens": 10439039.0,
      "step": 390
    },
    {
      "entropy": 0.771031835861504,
      "epoch": 0.7121397574273951,
      "grad_norm": 1.890625,
      "learning_rate": 7.980000000000002e-06,
      "loss": 0.7388,
      "mean_token_accuracy": 0.8006934400647878,
      "num_tokens": 10709549.0,
      "step": 400
    },
    {
      "epoch": 0.7121397574273951,
      "eval_chemistry_entropy": 0.8015456414222717,
      "eval_chemistry_loss": 0.782933235168457,
      "eval_chemistry_mean_token_accuracy": 0.7955212960243225,
      "eval_chemistry_num_tokens": 10709549.0,
      "eval_chemistry_runtime": 34.5635,
      "eval_chemistry_samples_per_second": 14.466,
      "eval_chemistry_steps_per_second": 3.617,
      "step": 400
    },
    {
      "epoch": 0.7121397574273951,
      "eval_physics_entropy": 0.6612502746582031,
      "eval_physics_loss": 0.6346089839935303,
      "eval_physics_mean_token_accuracy": 0.8259402804374695,
      "eval_physics_num_tokens": 10709549.0,
      "eval_physics_runtime": 41.2249,
      "eval_physics_samples_per_second": 12.129,
      "eval_physics_steps_per_second": 3.032,
      "step": 400
    },
    {
      "entropy": 0.7536357572302222,
      "epoch": 0.72994325136308,
      "grad_norm": 2.03125,
      "learning_rate": 8.18e-06,
      "loss": 0.7226,
      "mean_token_accuracy": 0.8046278320252895,
      "num_tokens": 10976011.0,
      "step": 410
    },
    {
      "entropy": 0.7753360515460372,
      "epoch": 0.7477467452987648,
      "grad_norm": 1.8203125,
      "learning_rate": 8.380000000000001e-06,
      "loss": 0.7386,
      "mean_token_accuracy": 0.8000223852694035,
      "num_tokens": 11245251.0,
      "step": 420
    },
    {
      "entropy": 0.7881368083879352,
      "epoch": 0.7655502392344498,
      "grad_norm": 2.140625,
      "learning_rate": 8.580000000000001e-06,
      "loss": 0.749,
      "mean_token_accuracy": 0.7968206465244293,
      "num_tokens": 11518514.0,
      "step": 430
    },
    {
      "entropy": 0.7809985620900989,
      "epoch": 0.7833537331701347,
      "grad_norm": 2.140625,
      "learning_rate": 8.78e-06,
      "loss": 0.7485,
      "mean_token_accuracy": 0.7970930390059948,
      "num_tokens": 11787117.0,
      "step": 440
    },
    {
      "entropy": 0.7746961690485478,
      "epoch": 0.8011572271058195,
      "grad_norm": 2.109375,
      "learning_rate": 8.98e-06,
      "loss": 0.7393,
      "mean_token_accuracy": 0.8000770397484303,
      "num_tokens": 12050993.0,
      "step": 450
    },
    {
      "entropy": 0.7727582400664688,
      "epoch": 0.8189607210415044,
      "grad_norm": 2.046875,
      "learning_rate": 9.180000000000002e-06,
      "loss": 0.7349,
      "mean_token_accuracy": 0.7991833999752999,
      "num_tokens": 12311688.0,
      "step": 460
    },
    {
      "entropy": 0.7381363824009896,
      "epoch": 0.8367642149771892,
      "grad_norm": 1.96875,
      "learning_rate": 9.38e-06,
      "loss": 0.7018,
      "mean_token_accuracy": 0.8056569200009107,
      "num_tokens": 12598408.0,
      "step": 470
    },
    {
      "entropy": 0.7458288084715605,
      "epoch": 0.8545677089128741,
      "grad_norm": 1.765625,
      "learning_rate": 9.58e-06,
      "loss": 0.7163,
      "mean_token_accuracy": 0.8052155710756779,
      "num_tokens": 12869642.0,
      "step": 480
    },
    {
      "entropy": 0.7388540361076593,
      "epoch": 0.8723712028485591,
      "grad_norm": 1.7578125,
      "learning_rate": 9.780000000000001e-06,
      "loss": 0.702,
      "mean_token_accuracy": 0.8063301011919976,
      "num_tokens": 13139555.0,
      "step": 490
    },
    {
      "entropy": 0.7744078584015369,
      "epoch": 0.8901746967842439,
      "grad_norm": 1.921875,
      "learning_rate": 9.980000000000001e-06,
      "loss": 0.7401,
      "mean_token_accuracy": 0.7977699730545282,
      "num_tokens": 13396545.0,
      "step": 500
    },
    {
      "epoch": 0.8901746967842439,
      "eval_chemistry_entropy": 0.7673972373008728,
      "eval_chemistry_loss": 0.7539037466049194,
      "eval_chemistry_mean_token_accuracy": 0.7999647192955017,
      "eval_chemistry_num_tokens": 13396545.0,
      "eval_chemistry_runtime": 34.5998,
      "eval_chemistry_samples_per_second": 14.451,
      "eval_chemistry_steps_per_second": 3.613,
      "step": 500
    },
    {
      "epoch": 0.8901746967842439,
      "eval_physics_entropy": 0.6430810136795044,
      "eval_physics_loss": 0.6303399205207825,
      "eval_physics_mean_token_accuracy": 0.8259163227081299,
      "eval_physics_num_tokens": 13396545.0,
      "eval_physics_runtime": 41.284,
      "eval_physics_samples_per_second": 12.111,
      "eval_physics_steps_per_second": 3.028,
      "step": 500
    },
    {
      "entropy": 0.7333443723618984,
      "epoch": 0.9079781907199288,
      "grad_norm": 1.625,
      "learning_rate": 1.018e-05,
      "loss": 0.6952,
      "mean_token_accuracy": 0.8094376768916846,
      "num_tokens": 13667857.0,
      "step": 510
    },
    {
      "entropy": 0.7600304771214723,
      "epoch": 0.9257816846556136,
      "grad_norm": 1.921875,
      "learning_rate": 1.038e-05,
      "loss": 0.7179,
      "mean_token_accuracy": 0.8037680003792047,
      "num_tokens": 13932771.0,
      "step": 520
    },
    {
      "entropy": 0.7434984143823385,
      "epoch": 0.9435851785912985,
      "grad_norm": 1.7578125,
      "learning_rate": 1.0580000000000002e-05,
      "loss": 0.7058,
      "mean_token_accuracy": 0.8062188472598791,
      "num_tokens": 14194632.0,
      "step": 530
    },
    {
      "entropy": 0.7611551772803068,
      "epoch": 0.9613886725269835,
      "grad_norm": 1.8203125,
      "learning_rate": 1.0780000000000002e-05,
      "loss": 0.7268,
      "mean_token_accuracy": 0.8000180393457412,
      "num_tokens": 14457650.0,
      "step": 540
    },
    {
      "entropy": 0.7252845833078027,
      "epoch": 0.9791921664626683,
      "grad_norm": 1.796875,
      "learning_rate": 1.0980000000000002e-05,
      "loss": 0.6969,
      "mean_token_accuracy": 0.8071364350616932,
      "num_tokens": 14724967.0,
      "step": 550
    },
    {
      "entropy": 0.7578858522698283,
      "epoch": 0.9969956603983532,
      "grad_norm": 1.8671875,
      "learning_rate": 1.1180000000000001e-05,
      "loss": 0.7274,
      "mean_token_accuracy": 0.8026934862136841,
      "num_tokens": 14993272.0,
      "step": 560
    },
    {
      "entropy": 0.6899186813062237,
      "epoch": 1.014242795148548,
      "grad_norm": 1.6484375,
      "learning_rate": 1.138e-05,
      "loss": 0.6595,
      "mean_token_accuracy": 0.8154077602970985,
      "num_tokens": 15254936.0,
      "step": 570
    },
    {
      "entropy": 0.7271509811282157,
      "epoch": 1.0320462890842328,
      "grad_norm": 2.03125,
      "learning_rate": 1.1580000000000001e-05,
      "loss": 0.6948,
      "mean_token_accuracy": 0.804929868131876,
      "num_tokens": 15515007.0,
      "step": 580
    },
    {
      "entropy": 0.6891195146366954,
      "epoch": 1.0498497830199176,
      "grad_norm": 1.875,
      "learning_rate": 1.178e-05,
      "loss": 0.6619,
      "mean_token_accuracy": 0.8155979864299298,
      "num_tokens": 15794203.0,
      "step": 590
    },
    {
      "entropy": 0.7109181759878993,
      "epoch": 1.0676532769556026,
      "grad_norm": 1.828125,
      "learning_rate": 1.198e-05,
      "loss": 0.6773,
      "mean_token_accuracy": 0.8111642371863127,
      "num_tokens": 16063664.0,
      "step": 600
    },
    {
      "epoch": 1.0676532769556026,
      "eval_chemistry_entropy": 0.7275329558849335,
      "eval_chemistry_loss": 0.7331510186195374,
      "eval_chemistry_mean_token_accuracy": 0.8037505540847778,
      "eval_chemistry_num_tokens": 16063664.0,
      "eval_chemistry_runtime": 34.6124,
      "eval_chemistry_samples_per_second": 14.446,
      "eval_chemistry_steps_per_second": 3.611,
      "step": 600
    },
    {
      "epoch": 1.0676532769556026,
      "eval_physics_entropy": 0.6278075368404389,
      "eval_physics_loss": 0.6282901763916016,
      "eval_physics_mean_token_accuracy": 0.825420051574707,
      "eval_physics_num_tokens": 16063664.0,
      "eval_physics_runtime": 41.2841,
      "eval_physics_samples_per_second": 12.111,
      "eval_physics_steps_per_second": 3.028,
      "step": 600
    },
    {
      "entropy": 0.6886052008718252,
      "epoch": 1.0854567708912874,
      "grad_norm": 1.9296875,
      "learning_rate": 1.218e-05,
      "loss": 0.664,
      "mean_token_accuracy": 0.8155276246368885,
      "num_tokens": 16338303.0,
      "step": 610
    },
    {
      "entropy": 0.7301610294729471,
      "epoch": 1.1032602648269723,
      "grad_norm": 2.03125,
      "learning_rate": 1.2380000000000002e-05,
      "loss": 0.69,
      "mean_token_accuracy": 0.8072479218244553,
      "num_tokens": 16606302.0,
      "step": 620
    },
    {
      "entropy": 0.7009955205023288,
      "epoch": 1.121063758762657,
      "grad_norm": 1.921875,
      "learning_rate": 1.2580000000000002e-05,
      "loss": 0.6662,
      "mean_token_accuracy": 0.8098525639623404,
      "num_tokens": 16877071.0,
      "step": 630
    },
    {
      "entropy": 0.6957393456250429,
      "epoch": 1.138867252698342,
      "grad_norm": 1.7421875,
      "learning_rate": 1.2780000000000001e-05,
      "loss": 0.6634,
      "mean_token_accuracy": 0.8121555853635073,
      "num_tokens": 17145775.0,
      "step": 640
    },
    {
      "entropy": 0.7081588903442025,
      "epoch": 1.156670746634027,
      "grad_norm": 1.828125,
      "learning_rate": 1.2980000000000001e-05,
      "loss": 0.6716,
      "mean_token_accuracy": 0.8104519549757242,
      "num_tokens": 17410775.0,
      "step": 650
    },
    {
      "entropy": 0.7330411653965712,
      "epoch": 1.1744742405697117,
      "grad_norm": 2.015625,
      "learning_rate": 1.3180000000000001e-05,
      "loss": 0.6949,
      "mean_token_accuracy": 0.8045349054038524,
      "num_tokens": 17677791.0,
      "step": 660
    },
    {
      "entropy": 0.7073222378268837,
      "epoch": 1.1922777345053968,
      "grad_norm": 1.8046875,
      "learning_rate": 1.3380000000000002e-05,
      "loss": 0.666,
      "mean_token_accuracy": 0.8106517169624567,
      "num_tokens": 17949441.0,
      "step": 670
    },
    {
      "entropy": 0.697087805531919,
      "epoch": 1.2100812284410816,
      "grad_norm": 1.8203125,
      "learning_rate": 1.3580000000000002e-05,
      "loss": 0.6644,
      "mean_token_accuracy": 0.8117371387779713,
      "num_tokens": 18217742.0,
      "step": 680
    },
    {
      "entropy": 0.7058945974335075,
      "epoch": 1.2278847223767664,
      "grad_norm": 1.9765625,
      "learning_rate": 1.378e-05,
      "loss": 0.6693,
      "mean_token_accuracy": 0.8107333112508058,
      "num_tokens": 18487769.0,
      "step": 690
    },
    {
      "entropy": 0.6924632461741567,
      "epoch": 1.2456882163124514,
      "grad_norm": 1.859375,
      "learning_rate": 1.398e-05,
      "loss": 0.6628,
      "mean_token_accuracy": 0.811711684986949,
      "num_tokens": 18760296.0,
      "step": 700
    },
    {
      "epoch": 1.2456882163124514,
      "eval_chemistry_entropy": 0.7344678246974945,
      "eval_chemistry_loss": 0.7194931507110596,
      "eval_chemistry_mean_token_accuracy": 0.8056631999015809,
      "eval_chemistry_num_tokens": 18760296.0,
      "eval_chemistry_runtime": 34.6183,
      "eval_chemistry_samples_per_second": 14.443,
      "eval_chemistry_steps_per_second": 3.611,
      "step": 700
    },
    {
      "epoch": 1.2456882163124514,
      "eval_physics_entropy": 0.6411541821956634,
      "eval_physics_loss": 0.6283924579620361,
      "eval_physics_mean_token_accuracy": 0.8252067227363586,
      "eval_physics_num_tokens": 18760296.0,
      "eval_physics_runtime": 41.2883,
      "eval_physics_samples_per_second": 12.11,
      "eval_physics_steps_per_second": 3.027,
      "step": 700
    },
    {
      "entropy": 0.6927230576053262,
      "epoch": 1.2634917102481362,
      "grad_norm": 1.6796875,
      "learning_rate": 1.418e-05,
      "loss": 0.6577,
      "mean_token_accuracy": 0.8144037745893001,
      "num_tokens": 19035733.0,
      "step": 710
    },
    {
      "entropy": 0.7126594386994839,
      "epoch": 1.281295204183821,
      "grad_norm": 1.78125,
      "learning_rate": 1.4380000000000001e-05,
      "loss": 0.681,
      "mean_token_accuracy": 0.80786834359169,
      "num_tokens": 19311432.0,
      "step": 720
    },
    {
      "entropy": 0.6811089560389518,
      "epoch": 1.299098698119506,
      "grad_norm": 1.6640625,
      "learning_rate": 1.4580000000000001e-05,
      "loss": 0.65,
      "mean_token_accuracy": 0.8166949838399887,
      "num_tokens": 19593016.0,
      "step": 730
    },
    {
      "entropy": 0.6871400082483887,
      "epoch": 1.3169021920551909,
      "grad_norm": 1.8359375,
      "learning_rate": 1.478e-05,
      "loss": 0.6507,
      "mean_token_accuracy": 0.8143899407237768,
      "num_tokens": 19860476.0,
      "step": 740
    },
    {
      "entropy": 0.6881214780732989,
      "epoch": 1.3347056859908757,
      "grad_norm": 1.9375,
      "learning_rate": 1.498e-05,
      "loss": 0.6517,
      "mean_token_accuracy": 0.8132891681045293,
      "num_tokens": 20124436.0,
      "step": 750
    },
    {
      "entropy": 0.6930879963561892,
      "epoch": 1.3525091799265607,
      "grad_norm": 1.7734375,
      "learning_rate": 1.5180000000000002e-05,
      "loss": 0.6664,
      "mean_token_accuracy": 0.8128124658018351,
      "num_tokens": 20388027.0,
      "step": 760
    },
    {
      "entropy": 0.6938662787899375,
      "epoch": 1.3703126738622455,
      "grad_norm": 1.8828125,
      "learning_rate": 1.5380000000000002e-05,
      "loss": 0.6592,
      "mean_token_accuracy": 0.812091863155365,
      "num_tokens": 20656076.0,
      "step": 770
    },
    {
      "entropy": 0.7249738791957497,
      "epoch": 1.3881161677979303,
      "grad_norm": 1.875,
      "learning_rate": 1.5580000000000003e-05,
      "loss": 0.6947,
      "mean_token_accuracy": 0.8059011038392783,
      "num_tokens": 20911633.0,
      "step": 780
    },
    {
      "entropy": 0.7112141275778413,
      "epoch": 1.4059196617336152,
      "grad_norm": 1.6640625,
      "learning_rate": 1.578e-05,
      "loss": 0.6839,
      "mean_token_accuracy": 0.8064889151602983,
      "num_tokens": 21176227.0,
      "step": 790
    },
    {
      "entropy": 0.6923007164150476,
      "epoch": 1.4237231556693,
      "grad_norm": 1.8828125,
      "learning_rate": 1.5980000000000003e-05,
      "loss": 0.6619,
      "mean_token_accuracy": 0.8126598935574293,
      "num_tokens": 21445242.0,
      "step": 800
    },
    {
      "epoch": 1.4237231556693,
      "eval_chemistry_entropy": 0.7049469864368438,
      "eval_chemistry_loss": 0.7038090229034424,
      "eval_chemistry_mean_token_accuracy": 0.8079238352775574,
      "eval_chemistry_num_tokens": 21445242.0,
      "eval_chemistry_runtime": 34.5889,
      "eval_chemistry_samples_per_second": 14.456,
      "eval_chemistry_steps_per_second": 3.614,
      "step": 800
    },
    {
      "epoch": 1.4237231556693,
      "eval_physics_entropy": 0.6302921390533447,
      "eval_physics_loss": 0.6273708343505859,
      "eval_physics_mean_token_accuracy": 0.824629822731018,
      "eval_physics_num_tokens": 21445242.0,
      "eval_physics_runtime": 41.2241,
      "eval_physics_samples_per_second": 12.129,
      "eval_physics_steps_per_second": 3.032,
      "step": 800
    },
    {
      "entropy": 0.6938302194699645,
      "epoch": 1.441526649604985,
      "grad_norm": 1.9765625,
      "learning_rate": 1.618e-05,
      "loss": 0.6643,
      "mean_token_accuracy": 0.8114070001989603,
      "num_tokens": 21718455.0,
      "step": 810
    },
    {
      "entropy": 0.6958711674436927,
      "epoch": 1.4593301435406698,
      "grad_norm": 1.8203125,
      "learning_rate": 1.638e-05,
      "loss": 0.6619,
      "mean_token_accuracy": 0.8139785166829825,
      "num_tokens": 21974076.0,
      "step": 820
    },
    {
      "entropy": 0.6980619359761476,
      "epoch": 1.4771336374763546,
      "grad_norm": 1.828125,
      "learning_rate": 1.658e-05,
      "loss": 0.6687,
      "mean_token_accuracy": 0.8110807090997696,
      "num_tokens": 22241599.0,
      "step": 830
    },
    {
      "entropy": 0.6977157359942794,
      "epoch": 1.4949371314120397,
      "grad_norm": 1.9921875,
      "learning_rate": 1.6780000000000002e-05,
      "loss": 0.6706,
      "mean_token_accuracy": 0.8098580442368984,
      "num_tokens": 22510725.0,
      "step": 840
    },
    {
      "entropy": 0.6765838697552681,
      "epoch": 1.5127406253477245,
      "grad_norm": 1.640625,
      "learning_rate": 1.698e-05,
      "loss": 0.6549,
      "mean_token_accuracy": 0.8153976205736398,
      "num_tokens": 22778514.0,
      "step": 850
    },
    {
      "entropy": 0.6760151350870729,
      "epoch": 1.5305441192834093,
      "grad_norm": 1.859375,
      "learning_rate": 1.718e-05,
      "loss": 0.6375,
      "mean_token_accuracy": 0.8156527172774076,
      "num_tokens": 23051230.0,
      "step": 860
    },
    {
      "entropy": 0.6908003646880388,
      "epoch": 1.5483476132190943,
      "grad_norm": 1.78125,
      "learning_rate": 1.7380000000000003e-05,
      "loss": 0.658,
      "mean_token_accuracy": 0.8122621558606624,
      "num_tokens": 23306327.0,
      "step": 870
    },
    {
      "entropy": 0.6964091071859002,
      "epoch": 1.5661511071547791,
      "grad_norm": 1.7578125,
      "learning_rate": 1.758e-05,
      "loss": 0.6597,
      "mean_token_accuracy": 0.811262060701847,
      "num_tokens": 23573127.0,
      "step": 880
    },
    {
      "entropy": 0.6847964959219098,
      "epoch": 1.583954601090464,
      "grad_norm": 1.5703125,
      "learning_rate": 1.7780000000000003e-05,
      "loss": 0.6532,
      "mean_token_accuracy": 0.8122437499463558,
      "num_tokens": 23841914.0,
      "step": 890
    },
    {
      "entropy": 0.6890485780313611,
      "epoch": 1.601758095026149,
      "grad_norm": 1.828125,
      "learning_rate": 1.798e-05,
      "loss": 0.6587,
      "mean_token_accuracy": 0.8129192098975182,
      "num_tokens": 24105114.0,
      "step": 900
    },
    {
      "epoch": 1.601758095026149,
      "eval_chemistry_entropy": 0.6863252129554749,
      "eval_chemistry_loss": 0.6930021643638611,
      "eval_chemistry_mean_token_accuracy": 0.8097150478363037,
      "eval_chemistry_num_tokens": 24105114.0,
      "eval_chemistry_runtime": 34.6005,
      "eval_chemistry_samples_per_second": 14.451,
      "eval_chemistry_steps_per_second": 3.613,
      "step": 900
    },
    {
      "epoch": 1.601758095026149,
      "eval_physics_entropy": 0.6244586195945739,
      "eval_physics_loss": 0.6287086606025696,
      "eval_physics_mean_token_accuracy": 0.8245933861732483,
      "eval_physics_num_tokens": 24105114.0,
      "eval_physics_runtime": 41.2389,
      "eval_physics_samples_per_second": 12.124,
      "eval_physics_steps_per_second": 3.031,
      "step": 900
    },
    {
      "entropy": 0.6585973870009184,
      "epoch": 1.6195615889618338,
      "grad_norm": 1.5703125,
      "learning_rate": 1.8180000000000002e-05,
      "loss": 0.6354,
      "mean_token_accuracy": 0.8186014011502266,
      "num_tokens": 24384008.0,
      "step": 910
    },
    {
      "entropy": 0.693634419888258,
      "epoch": 1.6373650828975186,
      "grad_norm": 1.78125,
      "learning_rate": 1.8380000000000004e-05,
      "loss": 0.6526,
      "mean_token_accuracy": 0.812073552981019,
      "num_tokens": 24638205.0,
      "step": 920
    },
    {
      "entropy": 0.6443359967321157,
      "epoch": 1.6551685768332036,
      "grad_norm": 1.640625,
      "learning_rate": 1.858e-05,
      "loss": 0.6138,
      "mean_token_accuracy": 0.8221632152795791,
      "num_tokens": 24910104.0,
      "step": 930
    },
    {
      "entropy": 0.6681255014613271,
      "epoch": 1.6729720707688884,
      "grad_norm": 1.8828125,
      "learning_rate": 1.878e-05,
      "loss": 0.6349,
      "mean_token_accuracy": 0.8162749476730824,
      "num_tokens": 25181306.0,
      "step": 940
    },
    {
      "entropy": 0.676227905228734,
      "epoch": 1.6907755647045732,
      "grad_norm": 1.75,
      "learning_rate": 1.898e-05,
      "loss": 0.6452,
      "mean_token_accuracy": 0.8150841273367405,
      "num_tokens": 25449138.0,
      "step": 950
    },
    {
      "entropy": 0.6844975538551807,
      "epoch": 1.7085790586402583,
      "grad_norm": 1.703125,
      "learning_rate": 1.918e-05,
      "loss": 0.6493,
      "mean_token_accuracy": 0.8142464257776737,
      "num_tokens": 25709030.0,
      "step": 960
    },
    {
      "entropy": 0.684933065995574,
      "epoch": 1.7263825525759429,
      "grad_norm": 1.828125,
      "learning_rate": 1.938e-05,
      "loss": 0.6539,
      "mean_token_accuracy": 0.8128440115600825,
      "num_tokens": 25972807.0,
      "step": 970
    },
    {
      "entropy": 0.6664420425891876,
      "epoch": 1.744186046511628,
      "grad_norm": 1.5234375,
      "learning_rate": 1.9580000000000002e-05,
      "loss": 0.632,
      "mean_token_accuracy": 0.8175933144986629,
      "num_tokens": 26244955.0,
      "step": 980
    },
    {
      "entropy": 0.639144916087389,
      "epoch": 1.761989540447313,
      "grad_norm": 1.515625,
      "learning_rate": 1.978e-05,
      "loss": 0.6137,
      "mean_token_accuracy": 0.8227979384362698,
      "num_tokens": 26515548.0,
      "step": 990
    },
    {
      "entropy": 0.6666794860735535,
      "epoch": 1.7797930343829975,
      "grad_norm": 1.578125,
      "learning_rate": 1.9980000000000002e-05,
      "loss": 0.6268,
      "mean_token_accuracy": 0.8173129990696907,
      "num_tokens": 26782376.0,
      "step": 1000
    },
    {
      "epoch": 1.7797930343829975,
      "eval_chemistry_entropy": 0.6736977963447571,
      "eval_chemistry_loss": 0.6848514676094055,
      "eval_chemistry_mean_token_accuracy": 0.8112705006599427,
      "eval_chemistry_num_tokens": 26782376.0,
      "eval_chemistry_runtime": 34.6023,
      "eval_chemistry_samples_per_second": 14.45,
      "eval_chemistry_steps_per_second": 3.612,
      "step": 1000
    },
    {
      "epoch": 1.7797930343829975,
      "eval_physics_entropy": 0.6228616139888763,
      "eval_physics_loss": 0.6303527355194092,
      "eval_physics_mean_token_accuracy": 0.8230352792739868,
      "eval_physics_num_tokens": 26782376.0,
      "eval_physics_runtime": 40.9503,
      "eval_physics_samples_per_second": 12.21,
      "eval_physics_steps_per_second": 3.052,
      "step": 1000
    },
    {
      "entropy": 0.6698539135977626,
      "epoch": 1.7975965283186826,
      "grad_norm": 1.6796875,
      "learning_rate": 1.9980000000000002e-05,
      "loss": 0.64,
      "mean_token_accuracy": 0.815592122823,
      "num_tokens": 27049648.0,
      "step": 1010
    },
    {
      "entropy": 0.6951070323586463,
      "epoch": 1.8154000222543676,
      "grad_norm": 1.6796875,
      "learning_rate": 1.995777777777778e-05,
      "loss": 0.6636,
      "mean_token_accuracy": 0.810206351801753,
      "num_tokens": 27315061.0,
      "step": 1020
    },
    {
      "entropy": 0.6561093555763364,
      "epoch": 1.8332035161900522,
      "grad_norm": 1.65625,
      "learning_rate": 1.9935555555555557e-05,
      "loss": 0.6193,
      "mean_token_accuracy": 0.8209650807082653,
      "num_tokens": 27587650.0,
      "step": 1030
    },
    {
      "entropy": 0.673011284135282,
      "epoch": 1.8510070101257372,
      "grad_norm": 1.4609375,
      "learning_rate": 1.9913333333333335e-05,
      "loss": 0.6476,
      "mean_token_accuracy": 0.8167991150170565,
      "num_tokens": 27855458.0,
      "step": 1040
    },
    {
      "entropy": 0.6599440133199096,
      "epoch": 1.868810504061422,
      "grad_norm": 1.578125,
      "learning_rate": 1.9891111111111112e-05,
      "loss": 0.6309,
      "mean_token_accuracy": 0.82068553827703,
      "num_tokens": 28127703.0,
      "step": 1050
    },
    {
      "entropy": 0.6387615172192455,
      "epoch": 1.8866139979971068,
      "grad_norm": 1.375,
      "learning_rate": 1.986888888888889e-05,
      "loss": 0.6027,
      "mean_token_accuracy": 0.8241243101656437,
      "num_tokens": 28395128.0,
      "step": 1060
    },
    {
      "entropy": 0.661963646300137,
      "epoch": 1.9044174919327919,
      "grad_norm": 1.5703125,
      "learning_rate": 1.9846666666666668e-05,
      "loss": 0.6311,
      "mean_token_accuracy": 0.817503708600998,
      "num_tokens": 28662344.0,
      "step": 1070
    },
    {
      "entropy": 0.6643290909007191,
      "epoch": 1.9222209858684767,
      "grad_norm": 1.6953125,
      "learning_rate": 1.9824444444444445e-05,
      "loss": 0.6355,
      "mean_token_accuracy": 0.8175244953483343,
      "num_tokens": 28928769.0,
      "step": 1080
    },
    {
      "entropy": 0.663678290694952,
      "epoch": 1.9400244798041615,
      "grad_norm": 1.8359375,
      "learning_rate": 1.9802222222222226e-05,
      "loss": 0.6318,
      "mean_token_accuracy": 0.8165966872125864,
      "num_tokens": 29186214.0,
      "step": 1090
    },
    {
      "entropy": 0.6729184232652188,
      "epoch": 1.9578279737398465,
      "grad_norm": 1.5078125,
      "learning_rate": 1.978e-05,
      "loss": 0.6437,
      "mean_token_accuracy": 0.8148789752274752,
      "num_tokens": 29446445.0,
      "step": 1100
    },
    {
      "epoch": 1.9578279737398465,
      "eval_chemistry_entropy": 0.667749724149704,
      "eval_chemistry_loss": 0.670352041721344,
      "eval_chemistry_mean_token_accuracy": 0.8144329328536988,
      "eval_chemistry_num_tokens": 29446445.0,
      "eval_chemistry_runtime": 34.5967,
      "eval_chemistry_samples_per_second": 14.452,
      "eval_chemistry_steps_per_second": 3.613,
      "step": 1100
    },
    {
      "epoch": 1.9578279737398465,
      "eval_physics_entropy": 0.6239387395381928,
      "eval_physics_loss": 0.6269068121910095,
      "eval_physics_mean_token_accuracy": 0.8240941405296326,
      "eval_physics_num_tokens": 29446445.0,
      "eval_physics_runtime": 41.2535,
      "eval_physics_samples_per_second": 12.12,
      "eval_physics_steps_per_second": 3.03,
      "step": 1100
    },
    {
      "entropy": 0.6582659302279353,
      "epoch": 1.9756314676755313,
      "grad_norm": 1.5234375,
      "learning_rate": 1.975777777777778e-05,
      "loss": 0.6207,
      "mean_token_accuracy": 0.8203308090567589,
      "num_tokens": 29711771.0,
      "step": 1110
    },
    {
      "entropy": 0.6604506926611066,
      "epoch": 1.9934349616112161,
      "grad_norm": 1.65625,
      "learning_rate": 1.9735555555555556e-05,
      "loss": 0.6317,
      "mean_token_accuracy": 0.8193308029323816,
      "num_tokens": 29979994.0,
      "step": 1120
    },
    {
      "entropy": 0.604119231431715,
      "epoch": 2.010682096361411,
      "grad_norm": 1.6953125,
      "learning_rate": 1.9713333333333337e-05,
      "loss": 0.5701,
      "mean_token_accuracy": 0.83187814950943,
      "num_tokens": 30240793.0,
      "step": 1130
    },
    {
      "entropy": 0.5444270031526685,
      "epoch": 2.028485590297096,
      "grad_norm": 1.671875,
      "learning_rate": 1.969111111111111e-05,
      "loss": 0.5188,
      "mean_token_accuracy": 0.8410136397928,
      "num_tokens": 30503988.0,
      "step": 1140
    },
    {
      "entropy": 0.5468037134036422,
      "epoch": 2.0462890842327806,
      "grad_norm": 1.8125,
      "learning_rate": 1.9668888888888892e-05,
      "loss": 0.5163,
      "mean_token_accuracy": 0.8412827380001545,
      "num_tokens": 30771841.0,
      "step": 1150
    },
    {
      "entropy": 0.5360686406493187,
      "epoch": 2.0640925781684656,
      "grad_norm": 1.8203125,
      "learning_rate": 1.9646666666666666e-05,
      "loss": 0.5083,
      "mean_token_accuracy": 0.8440209306776524,
      "num_tokens": 31045037.0,
      "step": 1160
    },
    {
      "entropy": 0.5755055965855718,
      "epoch": 2.0818960721041506,
      "grad_norm": 1.7890625,
      "learning_rate": 1.9624444444444447e-05,
      "loss": 0.5422,
      "mean_token_accuracy": 0.8334469784051179,
      "num_tokens": 31299575.0,
      "step": 1170
    },
    {
      "entropy": 0.5716140467673541,
      "epoch": 2.099699566039835,
      "grad_norm": 1.8671875,
      "learning_rate": 1.9602222222222225e-05,
      "loss": 0.5423,
      "mean_token_accuracy": 0.8336410760879517,
      "num_tokens": 31565963.0,
      "step": 1180
    },
    {
      "entropy": 0.5382599098607898,
      "epoch": 2.1175030599755202,
      "grad_norm": 1.6484375,
      "learning_rate": 1.9580000000000002e-05,
      "loss": 0.5168,
      "mean_token_accuracy": 0.8428799293935298,
      "num_tokens": 31834212.0,
      "step": 1190
    },
    {
      "entropy": 0.5634422322735191,
      "epoch": 2.1353065539112053,
      "grad_norm": 1.6953125,
      "learning_rate": 1.955777777777778e-05,
      "loss": 0.5327,
      "mean_token_accuracy": 0.8359647855162621,
      "num_tokens": 32103356.0,
      "step": 1200
    },
    {
      "epoch": 2.1353065539112053,
      "eval_chemistry_entropy": 0.5813326196670532,
      "eval_chemistry_loss": 0.6811516284942627,
      "eval_chemistry_mean_token_accuracy": 0.8141688752174377,
      "eval_chemistry_num_tokens": 32103356.0,
      "eval_chemistry_runtime": 34.6182,
      "eval_chemistry_samples_per_second": 14.443,
      "eval_chemistry_steps_per_second": 3.611,
      "step": 1200
    },
    {
      "epoch": 2.1353065539112053,
      "eval_physics_entropy": 0.5421837236881256,
      "eval_physics_loss": 0.6482248306274414,
      "eval_physics_mean_token_accuracy": 0.8212747263908386,
      "eval_physics_num_tokens": 32103356.0,
      "eval_physics_runtime": 41.2474,
      "eval_physics_samples_per_second": 12.122,
      "eval_physics_steps_per_second": 3.03,
      "step": 1200
    },
    {
      "entropy": 0.5539132442325354,
      "epoch": 2.15311004784689,
      "grad_norm": 1.6171875,
      "learning_rate": 1.9535555555555557e-05,
      "loss": 0.519,
      "mean_token_accuracy": 0.8395270679146052,
      "num_tokens": 32374830.0,
      "step": 1210
    },
    {
      "entropy": 0.5340574728325009,
      "epoch": 2.170913541782575,
      "grad_norm": 1.671875,
      "learning_rate": 1.9513333333333335e-05,
      "loss": 0.5083,
      "mean_token_accuracy": 0.8442022494971753,
      "num_tokens": 32643479.0,
      "step": 1220
    },
    {
      "entropy": 0.540911185927689,
      "epoch": 2.18871703571826,
      "grad_norm": 1.6953125,
      "learning_rate": 1.9491111111111113e-05,
      "loss": 0.5126,
      "mean_token_accuracy": 0.8427833508700132,
      "num_tokens": 32908761.0,
      "step": 1230
    },
    {
      "entropy": 0.5624004138633609,
      "epoch": 2.2065205296539445,
      "grad_norm": 1.5625,
      "learning_rate": 1.946888888888889e-05,
      "loss": 0.5288,
      "mean_token_accuracy": 0.8391110297292471,
      "num_tokens": 33166413.0,
      "step": 1240
    },
    {
      "entropy": 0.5559011250734329,
      "epoch": 2.2243240235896296,
      "grad_norm": 1.53125,
      "learning_rate": 1.9446666666666668e-05,
      "loss": 0.5314,
      "mean_token_accuracy": 0.8375788435339928,
      "num_tokens": 33432686.0,
      "step": 1250
    },
    {
      "entropy": 0.5524005191400647,
      "epoch": 2.242127517525314,
      "grad_norm": 1.8671875,
      "learning_rate": 1.9424444444444446e-05,
      "loss": 0.5252,
      "mean_token_accuracy": 0.8396644528955222,
      "num_tokens": 33693242.0,
      "step": 1260
    },
    {
      "entropy": 0.5266111794859171,
      "epoch": 2.259931011460999,
      "grad_norm": 1.6953125,
      "learning_rate": 1.9402222222222223e-05,
      "loss": 0.4966,
      "mean_token_accuracy": 0.847085589542985,
      "num_tokens": 33970820.0,
      "step": 1270
    },
    {
      "entropy": 0.5435060489922762,
      "epoch": 2.277734505396684,
      "grad_norm": 1.75,
      "learning_rate": 1.938e-05,
      "loss": 0.5191,
      "mean_token_accuracy": 0.8409085746854543,
      "num_tokens": 34243042.0,
      "step": 1280
    },
    {
      "entropy": 0.5497656201943755,
      "epoch": 2.295537999332369,
      "grad_norm": 1.6875,
      "learning_rate": 1.935777777777778e-05,
      "loss": 0.5245,
      "mean_token_accuracy": 0.8400060147047043,
      "num_tokens": 34515236.0,
      "step": 1290
    },
    {
      "entropy": 0.5251918986439705,
      "epoch": 2.313341493268054,
      "grad_norm": 1.6484375,
      "learning_rate": 1.9335555555555556e-05,
      "loss": 0.4979,
      "mean_token_accuracy": 0.848014684766531,
      "num_tokens": 34785068.0,
      "step": 1300
    },
    {
      "epoch": 2.313341493268054,
      "eval_chemistry_entropy": 0.5687024412155152,
      "eval_chemistry_loss": 0.6814095377922058,
      "eval_chemistry_mean_token_accuracy": 0.8144371585845948,
      "eval_chemistry_num_tokens": 34785068.0,
      "eval_chemistry_runtime": 34.6075,
      "eval_chemistry_samples_per_second": 14.448,
      "eval_chemistry_steps_per_second": 3.612,
      "step": 1300
    },
    {
      "epoch": 2.313341493268054,
      "eval_physics_entropy": 0.5373544907569885,
      "eval_physics_loss": 0.6523818373680115,
      "eval_physics_mean_token_accuracy": 0.8211935892105102,
      "eval_physics_num_tokens": 34785068.0,
      "eval_physics_runtime": 41.2549,
      "eval_physics_samples_per_second": 12.12,
      "eval_physics_steps_per_second": 3.03,
      "step": 1300
    },
    {
      "entropy": 0.5557622062042356,
      "epoch": 2.331144987203739,
      "grad_norm": 1.4921875,
      "learning_rate": 1.9313333333333334e-05,
      "loss": 0.5275,
      "mean_token_accuracy": 0.8374897003173828,
      "num_tokens": 35055845.0,
      "step": 1310
    },
    {
      "entropy": 0.5468069275841116,
      "epoch": 2.3489484811394234,
      "grad_norm": 1.6015625,
      "learning_rate": 1.9291111111111115e-05,
      "loss": 0.5189,
      "mean_token_accuracy": 0.8414008297026158,
      "num_tokens": 35327481.0,
      "step": 1320
    },
    {
      "entropy": 0.5578970231115818,
      "epoch": 2.3667519750751085,
      "grad_norm": 1.65625,
      "learning_rate": 1.926888888888889e-05,
      "loss": 0.5327,
      "mean_token_accuracy": 0.837207018584013,
      "num_tokens": 35596362.0,
      "step": 1330
    },
    {
      "entropy": 0.5461308034136891,
      "epoch": 2.3845554690107935,
      "grad_norm": 1.6328125,
      "learning_rate": 1.924666666666667e-05,
      "loss": 0.5144,
      "mean_token_accuracy": 0.8420327302068472,
      "num_tokens": 35869585.0,
      "step": 1340
    },
    {
      "entropy": 0.5518748119473458,
      "epoch": 2.402358962946478,
      "grad_norm": 1.765625,
      "learning_rate": 1.9224444444444444e-05,
      "loss": 0.5271,
      "mean_token_accuracy": 0.8386039260774851,
      "num_tokens": 36134072.0,
      "step": 1350
    },
    {
      "entropy": 0.5443382592871785,
      "epoch": 2.420162456882163,
      "grad_norm": 1.6015625,
      "learning_rate": 1.9202222222222225e-05,
      "loss": 0.5177,
      "mean_token_accuracy": 0.8425142876803875,
      "num_tokens": 36406263.0,
      "step": 1360
    },
    {
      "entropy": 0.5503683429211378,
      "epoch": 2.437965950817848,
      "grad_norm": 1.71875,
      "learning_rate": 1.918e-05,
      "loss": 0.5193,
      "mean_token_accuracy": 0.8414091795682908,
      "num_tokens": 36676822.0,
      "step": 1370
    },
    {
      "entropy": 0.5444278137758374,
      "epoch": 2.4557694447535328,
      "grad_norm": 1.8515625,
      "learning_rate": 1.915777777777778e-05,
      "loss": 0.5187,
      "mean_token_accuracy": 0.841403117403388,
      "num_tokens": 36942837.0,
      "step": 1380
    },
    {
      "entropy": 0.5489616293460131,
      "epoch": 2.473572938689218,
      "grad_norm": 1.8046875,
      "learning_rate": 1.9135555555555555e-05,
      "loss": 0.523,
      "mean_token_accuracy": 0.8393812999129295,
      "num_tokens": 37210791.0,
      "step": 1390
    },
    {
      "entropy": 0.5314275240525603,
      "epoch": 2.491376432624903,
      "grad_norm": 1.7265625,
      "learning_rate": 1.9113333333333336e-05,
      "loss": 0.5119,
      "mean_token_accuracy": 0.8442874882370234,
      "num_tokens": 37485822.0,
      "step": 1400
    },
    {
      "epoch": 2.491376432624903,
      "eval_chemistry_entropy": 0.5652910959720612,
      "eval_chemistry_loss": 0.6705389618873596,
      "eval_chemistry_mean_token_accuracy": 0.8175577726364136,
      "eval_chemistry_num_tokens": 37485822.0,
      "eval_chemistry_runtime": 34.6247,
      "eval_chemistry_samples_per_second": 14.441,
      "eval_chemistry_steps_per_second": 3.61,
      "step": 1400
    },
    {
      "epoch": 2.491376432624903,
      "eval_physics_entropy": 0.5328873710632325,
      "eval_physics_loss": 0.6484245657920837,
      "eval_physics_mean_token_accuracy": 0.8219989495277404,
      "eval_physics_num_tokens": 37485822.0,
      "eval_physics_runtime": 41.2499,
      "eval_physics_samples_per_second": 12.121,
      "eval_physics_steps_per_second": 3.03,
      "step": 1400
    },
    {
      "entropy": 0.5436741337180138,
      "epoch": 2.5091799265605874,
      "grad_norm": 1.65625,
      "learning_rate": 1.9091111111111113e-05,
      "loss": 0.5135,
      "mean_token_accuracy": 0.8424118865281344,
      "num_tokens": 37746150.0,
      "step": 1410
    },
    {
      "entropy": 0.5462065914645791,
      "epoch": 2.5269834204962724,
      "grad_norm": 1.609375,
      "learning_rate": 1.906888888888889e-05,
      "loss": 0.5158,
      "mean_token_accuracy": 0.8402331713587046,
      "num_tokens": 38016863.0,
      "step": 1420
    },
    {
      "entropy": 0.56711411960423,
      "epoch": 2.544786914431957,
      "grad_norm": 1.9765625,
      "learning_rate": 1.904666666666667e-05,
      "loss": 0.5349,
      "mean_token_accuracy": 0.8344442717730999,
      "num_tokens": 38277754.0,
      "step": 1430
    },
    {
      "entropy": 0.5571039380505681,
      "epoch": 2.562590408367642,
      "grad_norm": 1.734375,
      "learning_rate": 1.9024444444444446e-05,
      "loss": 0.529,
      "mean_token_accuracy": 0.8378622565418482,
      "num_tokens": 38542761.0,
      "step": 1440
    },
    {
      "entropy": 0.5426763394847512,
      "epoch": 2.580393902303327,
      "grad_norm": 1.796875,
      "learning_rate": 1.9002222222222224e-05,
      "loss": 0.5159,
      "mean_token_accuracy": 0.8411751441657543,
      "num_tokens": 38820999.0,
      "step": 1450
    },
    {
      "entropy": 0.5453944187611341,
      "epoch": 2.598197396239012,
      "grad_norm": 1.6953125,
      "learning_rate": 1.898e-05,
      "loss": 0.5238,
      "mean_token_accuracy": 0.8403888408094644,
      "num_tokens": 39080016.0,
      "step": 1460
    },
    {
      "entropy": 0.5451987128704786,
      "epoch": 2.6160008901746967,
      "grad_norm": 1.7734375,
      "learning_rate": 1.895777777777778e-05,
      "loss": 0.5123,
      "mean_token_accuracy": 0.8426350578665733,
      "num_tokens": 39345434.0,
      "step": 1470
    },
    {
      "entropy": 0.5478348048403859,
      "epoch": 2.6338043841103818,
      "grad_norm": 1.78125,
      "learning_rate": 1.8935555555555556e-05,
      "loss": 0.5212,
      "mean_token_accuracy": 0.8402178969234229,
      "num_tokens": 39610041.0,
      "step": 1480
    },
    {
      "entropy": 0.5249082477763295,
      "epoch": 2.6516078780460663,
      "grad_norm": 1.7890625,
      "learning_rate": 1.8913333333333334e-05,
      "loss": 0.4993,
      "mean_token_accuracy": 0.8461574167013168,
      "num_tokens": 39884185.0,
      "step": 1490
    },
    {
      "entropy": 0.5606990542262793,
      "epoch": 2.6694113719817514,
      "grad_norm": 1.953125,
      "learning_rate": 1.8891111111111115e-05,
      "loss": 0.533,
      "mean_token_accuracy": 0.8361463028937578,
      "num_tokens": 40143543.0,
      "step": 1500
    },
    {
      "epoch": 2.6694113719817514,
      "eval_chemistry_entropy": 0.5683894844055176,
      "eval_chemistry_loss": 0.6641016602516174,
      "eval_chemistry_mean_token_accuracy": 0.817899118423462,
      "eval_chemistry_num_tokens": 40143543.0,
      "eval_chemistry_runtime": 34.613,
      "eval_chemistry_samples_per_second": 14.445,
      "eval_chemistry_steps_per_second": 3.611,
      "step": 1500
    },
    {
      "epoch": 2.6694113719817514,
      "eval_physics_entropy": 0.5443970229625702,
      "eval_physics_loss": 0.6480051279067993,
      "eval_physics_mean_token_accuracy": 0.8217208013534546,
      "eval_physics_num_tokens": 40143543.0,
      "eval_physics_runtime": 40.9767,
      "eval_physics_samples_per_second": 12.202,
      "eval_physics_steps_per_second": 3.051,
      "step": 1500
    },
    {
      "entropy": 0.5430334324017168,
      "epoch": 2.6872148659174364,
      "grad_norm": 1.7421875,
      "learning_rate": 1.886888888888889e-05,
      "loss": 0.5152,
      "mean_token_accuracy": 0.8425797671079636,
      "num_tokens": 40406812.0,
      "step": 1510
    },
    {
      "entropy": 0.5591320911422372,
      "epoch": 2.7050183598531214,
      "grad_norm": 1.8359375,
      "learning_rate": 1.884666666666667e-05,
      "loss": 0.5348,
      "mean_token_accuracy": 0.8362568471580744,
      "num_tokens": 40668543.0,
      "step": 1520
    },
    {
      "entropy": 0.5530628427863121,
      "epoch": 2.722821853788806,
      "grad_norm": 1.6953125,
      "learning_rate": 1.8824444444444445e-05,
      "loss": 0.518,
      "mean_token_accuracy": 0.8411050379276276,
      "num_tokens": 40942603.0,
      "step": 1530
    },
    {
      "entropy": 0.556857710890472,
      "epoch": 2.740625347724491,
      "grad_norm": 1.6796875,
      "learning_rate": 1.8802222222222226e-05,
      "loss": 0.5243,
      "mean_token_accuracy": 0.838795681297779,
      "num_tokens": 41208093.0,
      "step": 1540
    },
    {
      "entropy": 0.5422420864924788,
      "epoch": 2.7584288416601757,
      "grad_norm": 1.609375,
      "learning_rate": 1.878e-05,
      "loss": 0.5204,
      "mean_token_accuracy": 0.8388970494270325,
      "num_tokens": 41477562.0,
      "step": 1550
    },
    {
      "entropy": 0.5319694015197456,
      "epoch": 2.7762323355958607,
      "grad_norm": 1.8359375,
      "learning_rate": 1.875777777777778e-05,
      "loss": 0.5078,
      "mean_token_accuracy": 0.8438673295080662,
      "num_tokens": 41746571.0,
      "step": 1560
    },
    {
      "entropy": 0.5427202990278601,
      "epoch": 2.7940358295315457,
      "grad_norm": 1.703125,
      "learning_rate": 1.873555555555556e-05,
      "loss": 0.5141,
      "mean_token_accuracy": 0.842416912689805,
      "num_tokens": 42011633.0,
      "step": 1570
    },
    {
      "entropy": 0.5518008042126894,
      "epoch": 2.8118393234672303,
      "grad_norm": 1.546875,
      "learning_rate": 1.8713333333333336e-05,
      "loss": 0.5238,
      "mean_token_accuracy": 0.8391041114926339,
      "num_tokens": 42283087.0,
      "step": 1580
    },
    {
      "entropy": 0.5269314305856824,
      "epoch": 2.8296428174029153,
      "grad_norm": 1.4375,
      "learning_rate": 1.8691111111111114e-05,
      "loss": 0.5,
      "mean_token_accuracy": 0.8462934419512749,
      "num_tokens": 42561534.0,
      "step": 1590
    },
    {
      "entropy": 0.54100974611938,
      "epoch": 2.8474463113386,
      "grad_norm": 1.6328125,
      "learning_rate": 1.866888888888889e-05,
      "loss": 0.5206,
      "mean_token_accuracy": 0.8429192952811718,
      "num_tokens": 42834266.0,
      "step": 1600
    },
    {
      "epoch": 2.8474463113386,
      "eval_chemistry_entropy": 0.5678292090892791,
      "eval_chemistry_loss": 0.6582387089729309,
      "eval_chemistry_mean_token_accuracy": 0.8187746982574463,
      "eval_chemistry_num_tokens": 42834266.0,
      "eval_chemistry_runtime": 34.6036,
      "eval_chemistry_samples_per_second": 14.449,
      "eval_chemistry_steps_per_second": 3.612,
      "step": 1600
    },
    {
      "epoch": 2.8474463113386,
      "eval_physics_entropy": 0.5397046103477477,
      "eval_physics_loss": 0.6487213969230652,
      "eval_physics_mean_token_accuracy": 0.8217604222297669,
      "eval_physics_num_tokens": 42834266.0,
      "eval_physics_runtime": 41.2474,
      "eval_physics_samples_per_second": 12.122,
      "eval_physics_steps_per_second": 3.03,
      "step": 1600
    },
    {
      "entropy": 0.5642041997984052,
      "epoch": 2.865249805274285,
      "grad_norm": 1.796875,
      "learning_rate": 1.864666666666667e-05,
      "loss": 0.5371,
      "mean_token_accuracy": 0.8362252287566662,
      "num_tokens": 43100513.0,
      "step": 1610
    },
    {
      "entropy": 0.553196731954813,
      "epoch": 2.88305329920997,
      "grad_norm": 1.65625,
      "learning_rate": 1.8624444444444446e-05,
      "loss": 0.5214,
      "mean_token_accuracy": 0.8403059497475625,
      "num_tokens": 43375632.0,
      "step": 1620
    },
    {
      "entropy": 0.5369223712012172,
      "epoch": 2.900856793145655,
      "grad_norm": 1.640625,
      "learning_rate": 1.8602222222222224e-05,
      "loss": 0.5167,
      "mean_token_accuracy": 0.8430781438946724,
      "num_tokens": 43644341.0,
      "step": 1630
    },
    {
      "entropy": 0.5357730442658066,
      "epoch": 2.9186602870813396,
      "grad_norm": 1.78125,
      "learning_rate": 1.858e-05,
      "loss": 0.5061,
      "mean_token_accuracy": 0.8430858284235001,
      "num_tokens": 43917580.0,
      "step": 1640
    },
    {
      "entropy": 0.5374868368729949,
      "epoch": 2.9364637810170247,
      "grad_norm": 1.7578125,
      "learning_rate": 1.855777777777778e-05,
      "loss": 0.5116,
      "mean_token_accuracy": 0.8447067823261023,
      "num_tokens": 44177325.0,
      "step": 1650
    },
    {
      "entropy": 0.5514863247051836,
      "epoch": 2.9542672749527092,
      "grad_norm": 1.6953125,
      "learning_rate": 1.8535555555555557e-05,
      "loss": 0.5211,
      "mean_token_accuracy": 0.8395764529705048,
      "num_tokens": 44442947.0,
      "step": 1660
    },
    {
      "entropy": 0.5531215896829963,
      "epoch": 2.9720707688883943,
      "grad_norm": 1.8515625,
      "learning_rate": 1.8513333333333335e-05,
      "loss": 0.5287,
      "mean_token_accuracy": 0.8382051270455122,
      "num_tokens": 44700601.0,
      "step": 1670
    },
    {
      "entropy": 0.5407275201752781,
      "epoch": 2.9898742628240793,
      "grad_norm": 1.59375,
      "learning_rate": 1.8491111111111112e-05,
      "loss": 0.51,
      "mean_token_accuracy": 0.844194658845663,
      "num_tokens": 44971479.0,
      "step": 1680
    },
    {
      "entropy": 0.49610343229386117,
      "epoch": 3.007121397574274,
      "grad_norm": 1.65625,
      "learning_rate": 1.846888888888889e-05,
      "loss": 0.4379,
      "mean_token_accuracy": 0.8622554202233591,
      "num_tokens": 45231796.0,
      "step": 1690
    },
    {
      "entropy": 0.3639660466462374,
      "epoch": 3.0249248915099587,
      "grad_norm": 2.171875,
      "learning_rate": 1.8446666666666667e-05,
      "loss": 0.3489,
      "mean_token_accuracy": 0.8842393532395363,
      "num_tokens": 45500317.0,
      "step": 1700
    },
    {
      "epoch": 3.0249248915099587,
      "eval_chemistry_entropy": 0.40514632713794707,
      "eval_chemistry_loss": 0.756369411945343,
      "eval_chemistry_mean_token_accuracy": 0.815176130771637,
      "eval_chemistry_num_tokens": 45500317.0,
      "eval_chemistry_runtime": 34.6159,
      "eval_chemistry_samples_per_second": 14.444,
      "eval_chemistry_steps_per_second": 3.611,
      "step": 1700
    },
    {
      "epoch": 3.0249248915099587,
      "eval_physics_entropy": 0.4021349341869354,
      "eval_physics_loss": 0.7458917498588562,
      "eval_physics_mean_token_accuracy": 0.8158441076278686,
      "eval_physics_num_tokens": 45500317.0,
      "eval_physics_runtime": 41.2473,
      "eval_physics_samples_per_second": 12.122,
      "eval_physics_steps_per_second": 3.031,
      "step": 1700
    },
    {
      "entropy": 0.35829943269491193,
      "epoch": 3.0427283854456437,
      "grad_norm": 2.125,
      "learning_rate": 1.842444444444445e-05,
      "loss": 0.3155,
      "mean_token_accuracy": 0.8941227950155735,
      "num_tokens": 45762680.0,
      "step": 1710
    },
    {
      "entropy": 0.3386535131372511,
      "epoch": 3.0605318793813288,
      "grad_norm": 2.296875,
      "learning_rate": 1.8402222222222223e-05,
      "loss": 0.3128,
      "mean_token_accuracy": 0.8955158289521933,
      "num_tokens": 46029923.0,
      "step": 1720
    },
    {
      "entropy": 0.3458150915801525,
      "epoch": 3.0783353733170133,
      "grad_norm": 2.078125,
      "learning_rate": 1.8380000000000004e-05,
      "loss": 0.3192,
      "mean_token_accuracy": 0.8931533485651016,
      "num_tokens": 46305156.0,
      "step": 1730
    },
    {
      "entropy": 0.3377646486274898,
      "epoch": 3.0961388672526984,
      "grad_norm": 2.1875,
      "learning_rate": 1.8357777777777778e-05,
      "loss": 0.3127,
      "mean_token_accuracy": 0.8952449705451727,
      "num_tokens": 46582874.0,
      "step": 1740
    },
    {
      "entropy": 0.35565020963549615,
      "epoch": 3.1139423611883834,
      "grad_norm": 2.59375,
      "learning_rate": 1.833555555555556e-05,
      "loss": 0.327,
      "mean_token_accuracy": 0.8901556842029095,
      "num_tokens": 46850175.0,
      "step": 1750
    },
    {
      "entropy": 0.3427689325064421,
      "epoch": 3.131745855124068,
      "grad_norm": 2.0625,
      "learning_rate": 1.8313333333333333e-05,
      "loss": 0.3146,
      "mean_token_accuracy": 0.8948938645422458,
      "num_tokens": 47125899.0,
      "step": 1760
    },
    {
      "entropy": 0.345877396594733,
      "epoch": 3.149549349059753,
      "grad_norm": 2.421875,
      "learning_rate": 1.8291111111111114e-05,
      "loss": 0.3207,
      "mean_token_accuracy": 0.8928927041590213,
      "num_tokens": 47390194.0,
      "step": 1770
    },
    {
      "entropy": 0.3544696198776364,
      "epoch": 3.167352842995438,
      "grad_norm": 2.203125,
      "learning_rate": 1.8268888888888888e-05,
      "loss": 0.324,
      "mean_token_accuracy": 0.8916583750396967,
      "num_tokens": 47654784.0,
      "step": 1780
    },
    {
      "entropy": 0.345364089962095,
      "epoch": 3.1851563369311227,
      "grad_norm": 2.25,
      "learning_rate": 1.824666666666667e-05,
      "loss": 0.32,
      "mean_token_accuracy": 0.8935606256127357,
      "num_tokens": 47919400.0,
      "step": 1790
    },
    {
      "entropy": 0.3512003905139863,
      "epoch": 3.2029598308668077,
      "grad_norm": 2.28125,
      "learning_rate": 1.8224444444444447e-05,
      "loss": 0.3262,
      "mean_token_accuracy": 0.8913596641272307,
      "num_tokens": 48187796.0,
      "step": 1800
    },
    {
      "epoch": 3.2029598308668077,
      "eval_chemistry_entropy": 0.39868619096279145,
      "eval_chemistry_loss": 0.7814938426017761,
      "eval_chemistry_mean_token_accuracy": 0.8135026731491088,
      "eval_chemistry_num_tokens": 48187796.0,
      "eval_chemistry_runtime": 34.5848,
      "eval_chemistry_samples_per_second": 14.457,
      "eval_chemistry_steps_per_second": 3.614,
      "step": 1800
    },
    {
      "epoch": 3.2029598308668077,
      "eval_physics_entropy": 0.38585546863079073,
      "eval_physics_loss": 0.785114049911499,
      "eval_physics_mean_token_accuracy": 0.8136380562782287,
      "eval_physics_num_tokens": 48187796.0,
      "eval_physics_runtime": 41.2383,
      "eval_physics_samples_per_second": 12.125,
      "eval_physics_steps_per_second": 3.031,
      "step": 1800
    },
    {
      "entropy": 0.34991011349484324,
      "epoch": 3.2207633248024923,
      "grad_norm": 2.21875,
      "learning_rate": 1.8202222222222225e-05,
      "loss": 0.3223,
      "mean_token_accuracy": 0.8920448690652847,
      "num_tokens": 48457428.0,
      "step": 1810
    },
    {
      "entropy": 0.3759195095859468,
      "epoch": 3.2385668187381773,
      "grad_norm": 2.359375,
      "learning_rate": 1.8180000000000002e-05,
      "loss": 0.341,
      "mean_token_accuracy": 0.8847057934850454,
      "num_tokens": 48716417.0,
      "step": 1820
    },
    {
      "entropy": 0.3503100727684796,
      "epoch": 3.2563703126738623,
      "grad_norm": 2.234375,
      "learning_rate": 1.815777777777778e-05,
      "loss": 0.3211,
      "mean_token_accuracy": 0.8925649128854275,
      "num_tokens": 48977306.0,
      "step": 1830
    },
    {
      "entropy": 0.34978028303012254,
      "epoch": 3.274173806609547,
      "grad_norm": 2.09375,
      "learning_rate": 1.8135555555555557e-05,
      "loss": 0.3206,
      "mean_token_accuracy": 0.8920259241014719,
      "num_tokens": 49249667.0,
      "step": 1840
    },
    {
      "entropy": 0.35568869626149535,
      "epoch": 3.291977300545232,
      "grad_norm": 2.109375,
      "learning_rate": 1.8113333333333335e-05,
      "loss": 0.3283,
      "mean_token_accuracy": 0.8901085384190083,
      "num_tokens": 49520900.0,
      "step": 1850
    },
    {
      "entropy": 0.35994688216596843,
      "epoch": 3.309780794480917,
      "grad_norm": 2.34375,
      "learning_rate": 1.8091111111111113e-05,
      "loss": 0.3309,
      "mean_token_accuracy": 0.8887050859630108,
      "num_tokens": 49786411.0,
      "step": 1860
    },
    {
      "entropy": 0.34956529131159186,
      "epoch": 3.3275842884166016,
      "grad_norm": 2.140625,
      "learning_rate": 1.806888888888889e-05,
      "loss": 0.3194,
      "mean_token_accuracy": 0.8924664918333292,
      "num_tokens": 50056207.0,
      "step": 1870
    },
    {
      "entropy": 0.35508423103019593,
      "epoch": 3.3453877823522866,
      "grad_norm": 2.078125,
      "learning_rate": 1.8046666666666668e-05,
      "loss": 0.3273,
      "mean_token_accuracy": 0.8904457587748766,
      "num_tokens": 50330876.0,
      "step": 1880
    },
    {
      "entropy": 0.35782372783869504,
      "epoch": 3.3631912762879717,
      "grad_norm": 2.4375,
      "learning_rate": 1.8024444444444445e-05,
      "loss": 0.3313,
      "mean_token_accuracy": 0.8893947552889585,
      "num_tokens": 50602004.0,
      "step": 1890
    },
    {
      "entropy": 0.36222805520519613,
      "epoch": 3.3809947702236562,
      "grad_norm": 2.203125,
      "learning_rate": 1.8002222222222223e-05,
      "loss": 0.3328,
      "mean_token_accuracy": 0.8882043663412332,
      "num_tokens": 50873444.0,
      "step": 1900
    },
    {
      "epoch": 3.3809947702236562,
      "eval_chemistry_entropy": 0.3968628585338593,
      "eval_chemistry_loss": 0.7823912501335144,
      "eval_chemistry_mean_token_accuracy": 0.8138084063529968,
      "eval_chemistry_num_tokens": 50873444.0,
      "eval_chemistry_runtime": 34.5801,
      "eval_chemistry_samples_per_second": 14.459,
      "eval_chemistry_steps_per_second": 3.615,
      "step": 1900
    },
    {
      "epoch": 3.3809947702236562,
      "eval_physics_entropy": 0.3825342948436737,
      "eval_physics_loss": 0.789954423904419,
      "eval_physics_mean_token_accuracy": 0.8131929535865784,
      "eval_physics_num_tokens": 50873444.0,
      "eval_physics_runtime": 41.2311,
      "eval_physics_samples_per_second": 12.127,
      "eval_physics_steps_per_second": 3.032,
      "step": 1900
    },
    {
      "entropy": 0.3620292558334768,
      "epoch": 3.3987982641593413,
      "grad_norm": 2.28125,
      "learning_rate": 1.798e-05,
      "loss": 0.3342,
      "mean_token_accuracy": 0.8878430653363466,
      "num_tokens": 51141077.0,
      "step": 1910
    },
    {
      "entropy": 0.3582372093573213,
      "epoch": 3.4166017580950263,
      "grad_norm": 2.375,
      "learning_rate": 1.7957777777777778e-05,
      "loss": 0.3297,
      "mean_token_accuracy": 0.8890867009758949,
      "num_tokens": 51406483.0,
      "step": 1920
    },
    {
      "entropy": 0.3618839686736465,
      "epoch": 3.434405252030711,
      "grad_norm": 2.4375,
      "learning_rate": 1.7935555555555556e-05,
      "loss": 0.3301,
      "mean_token_accuracy": 0.8890988543629647,
      "num_tokens": 51668458.0,
      "step": 1930
    },
    {
      "entropy": 0.36373381447046993,
      "epoch": 3.452208745966396,
      "grad_norm": 2.484375,
      "learning_rate": 1.7913333333333337e-05,
      "loss": 0.3361,
      "mean_token_accuracy": 0.8878721829503775,
      "num_tokens": 51928199.0,
      "step": 1940
    },
    {
      "entropy": 0.35968201477080586,
      "epoch": 3.470012239902081,
      "grad_norm": 2.21875,
      "learning_rate": 1.789111111111111e-05,
      "loss": 0.3309,
      "mean_token_accuracy": 0.8883917711675167,
      "num_tokens": 52199244.0,
      "step": 1950
    },
    {
      "entropy": 0.35602816781029106,
      "epoch": 3.4878157338377656,
      "grad_norm": 2.375,
      "learning_rate": 1.7868888888888892e-05,
      "loss": 0.3274,
      "mean_token_accuracy": 0.8896724134683609,
      "num_tokens": 52473476.0,
      "step": 1960
    },
    {
      "entropy": 0.3581710479222238,
      "epoch": 3.5056192277734506,
      "grad_norm": 1.9921875,
      "learning_rate": 1.7846666666666666e-05,
      "loss": 0.3307,
      "mean_token_accuracy": 0.8886904548853636,
      "num_tokens": 52743275.0,
      "step": 1970
    },
    {
      "entropy": 0.3580263523384929,
      "epoch": 3.523422721709135,
      "grad_norm": 2.171875,
      "learning_rate": 1.7824444444444447e-05,
      "loss": 0.328,
      "mean_token_accuracy": 0.8894063163548708,
      "num_tokens": 53014827.0,
      "step": 1980
    },
    {
      "entropy": 0.3583196923136711,
      "epoch": 3.54122621564482,
      "grad_norm": 2.46875,
      "learning_rate": 1.780222222222222e-05,
      "loss": 0.3328,
      "mean_token_accuracy": 0.8892181769013405,
      "num_tokens": 53285577.0,
      "step": 1990
    },
    {
      "entropy": 0.35019520940259097,
      "epoch": 3.5590297095805052,
      "grad_norm": 2.359375,
      "learning_rate": 1.7780000000000003e-05,
      "loss": 0.3253,
      "mean_token_accuracy": 0.8909101441502572,
      "num_tokens": 53553391.0,
      "step": 2000
    },
    {
      "epoch": 3.5590297095805052,
      "eval_chemistry_entropy": 0.40147681057453155,
      "eval_chemistry_loss": 0.778447151184082,
      "eval_chemistry_mean_token_accuracy": 0.8141673741340637,
      "eval_chemistry_num_tokens": 53553391.0,
      "eval_chemistry_runtime": 34.6216,
      "eval_chemistry_samples_per_second": 14.442,
      "eval_chemistry_steps_per_second": 3.61,
      "step": 2000
    },
    {
      "epoch": 3.5590297095805052,
      "eval_physics_entropy": 0.3869015793800354,
      "eval_physics_loss": 0.7880970239639282,
      "eval_physics_mean_token_accuracy": 0.813429744720459,
      "eval_physics_num_tokens": 53553391.0,
      "eval_physics_runtime": 40.9274,
      "eval_physics_samples_per_second": 12.217,
      "eval_physics_steps_per_second": 3.054,
      "step": 2000
    },
    {
      "entropy": 0.3586632375605404,
      "epoch": 3.5768332035161903,
      "grad_norm": 2.28125,
      "learning_rate": 1.7757777777777777e-05,
      "loss": 0.3297,
      "mean_token_accuracy": 0.8897864270955325,
      "num_tokens": 53826074.0,
      "step": 2010
    },
    {
      "entropy": 0.3541630276478827,
      "epoch": 3.594636697451875,
      "grad_norm": 2.375,
      "learning_rate": 1.7735555555555558e-05,
      "loss": 0.3274,
      "mean_token_accuracy": 0.8892519008368254,
      "num_tokens": 54090804.0,
      "step": 2020
    },
    {
      "entropy": 0.34958077147603034,
      "epoch": 3.61244019138756,
      "grad_norm": 2.296875,
      "learning_rate": 1.7713333333333335e-05,
      "loss": 0.3272,
      "mean_token_accuracy": 0.890841381624341,
      "num_tokens": 54350633.0,
      "step": 2030
    },
    {
      "entropy": 0.3693890837952495,
      "epoch": 3.6302436853232445,
      "grad_norm": 2.03125,
      "learning_rate": 1.7691111111111113e-05,
      "loss": 0.3406,
      "mean_token_accuracy": 0.885808477550745,
      "num_tokens": 54623090.0,
      "step": 2040
    },
    {
      "entropy": 0.3497920027002692,
      "epoch": 3.6480471792589295,
      "grad_norm": 2.1875,
      "learning_rate": 1.766888888888889e-05,
      "loss": 0.3254,
      "mean_token_accuracy": 0.891067099198699,
      "num_tokens": 54891961.0,
      "step": 2050
    },
    {
      "entropy": 0.35692696580663325,
      "epoch": 3.6658506731946146,
      "grad_norm": 2.28125,
      "learning_rate": 1.7646666666666668e-05,
      "loss": 0.332,
      "mean_token_accuracy": 0.8886408120393753,
      "num_tokens": 55160443.0,
      "step": 2060
    },
    {
      "entropy": 0.3660690987482667,
      "epoch": 3.6836541671302996,
      "grad_norm": 2.484375,
      "learning_rate": 1.7624444444444446e-05,
      "loss": 0.3363,
      "mean_token_accuracy": 0.8873372543603182,
      "num_tokens": 55425148.0,
      "step": 2070
    },
    {
      "entropy": 0.361936164367944,
      "epoch": 3.701457661065984,
      "grad_norm": 2.25,
      "learning_rate": 1.7602222222222223e-05,
      "loss": 0.3347,
      "mean_token_accuracy": 0.8881115224212408,
      "num_tokens": 55701877.0,
      "step": 2080
    },
    {
      "entropy": 0.36633992604911325,
      "epoch": 3.719261155001669,
      "grad_norm": 2.28125,
      "learning_rate": 1.758e-05,
      "loss": 0.3417,
      "mean_token_accuracy": 0.8862893104553222,
      "num_tokens": 55969756.0,
      "step": 2090
    },
    {
      "entropy": 0.35766666773706673,
      "epoch": 3.737064648937354,
      "grad_norm": 2.203125,
      "learning_rate": 1.755777777777778e-05,
      "loss": 0.3286,
      "mean_token_accuracy": 0.8889819283038378,
      "num_tokens": 56238576.0,
      "step": 2100
    },
    {
      "epoch": 3.737064648937354,
      "eval_chemistry_entropy": 0.3998841570615768,
      "eval_chemistry_loss": 0.7812446355819702,
      "eval_chemistry_mean_token_accuracy": 0.8146659145355225,
      "eval_chemistry_num_tokens": 56238576.0,
      "eval_chemistry_runtime": 34.6095,
      "eval_chemistry_samples_per_second": 14.447,
      "eval_chemistry_steps_per_second": 3.612,
      "step": 2100
    },
    {
      "epoch": 3.737064648937354,
      "eval_physics_entropy": 0.3837117667198181,
      "eval_physics_loss": 0.7952807545661926,
      "eval_physics_mean_token_accuracy": 0.8123790616989136,
      "eval_physics_num_tokens": 56238576.0,
      "eval_physics_runtime": 41.2513,
      "eval_physics_samples_per_second": 12.121,
      "eval_physics_steps_per_second": 3.03,
      "step": 2100
    },
    {
      "entropy": 0.3616418926976621,
      "epoch": 3.754868142873039,
      "grad_norm": 2.453125,
      "learning_rate": 1.7535555555555556e-05,
      "loss": 0.3361,
      "mean_token_accuracy": 0.8876710329204798,
      "num_tokens": 56497905.0,
      "step": 2110
    },
    {
      "entropy": 0.36285804882645606,
      "epoch": 3.772671636808724,
      "grad_norm": 2.625,
      "learning_rate": 1.7513333333333334e-05,
      "loss": 0.3358,
      "mean_token_accuracy": 0.8870962157845497,
      "num_tokens": 56757407.0,
      "step": 2120
    },
    {
      "entropy": 0.3556675191037357,
      "epoch": 3.7904751307444084,
      "grad_norm": 2.234375,
      "learning_rate": 1.749111111111111e-05,
      "loss": 0.3304,
      "mean_token_accuracy": 0.8892902947962285,
      "num_tokens": 57019213.0,
      "step": 2130
    },
    {
      "entropy": 0.35712740384042263,
      "epoch": 3.8082786246800935,
      "grad_norm": 2.234375,
      "learning_rate": 1.746888888888889e-05,
      "loss": 0.3322,
      "mean_token_accuracy": 0.8880893222987651,
      "num_tokens": 57274629.0,
      "step": 2140
    },
    {
      "entropy": 0.36487481808289884,
      "epoch": 3.826082118615778,
      "grad_norm": 2.265625,
      "learning_rate": 1.7446666666666667e-05,
      "loss": 0.3348,
      "mean_token_accuracy": 0.8869650844484568,
      "num_tokens": 57549791.0,
      "step": 2150
    },
    {
      "entropy": 0.354267621319741,
      "epoch": 3.843885612551463,
      "grad_norm": 2.40625,
      "learning_rate": 1.7424444444444444e-05,
      "loss": 0.3293,
      "mean_token_accuracy": 0.8896507594734431,
      "num_tokens": 57815454.0,
      "step": 2160
    },
    {
      "entropy": 0.36807344257831576,
      "epoch": 3.861689106487148,
      "grad_norm": 2.21875,
      "learning_rate": 1.7402222222222222e-05,
      "loss": 0.3403,
      "mean_token_accuracy": 0.8864047069102525,
      "num_tokens": 58076430.0,
      "step": 2170
    },
    {
      "entropy": 0.3607250331901014,
      "epoch": 3.879492600422833,
      "grad_norm": 2.515625,
      "learning_rate": 1.7380000000000003e-05,
      "loss": 0.3365,
      "mean_token_accuracy": 0.8868528883904219,
      "num_tokens": 58346307.0,
      "step": 2180
    },
    {
      "entropy": 0.36714982697740195,
      "epoch": 3.8972960943585178,
      "grad_norm": 2.296875,
      "learning_rate": 1.735777777777778e-05,
      "loss": 0.3409,
      "mean_token_accuracy": 0.8862496245652437,
      "num_tokens": 58621624.0,
      "step": 2190
    },
    {
      "entropy": 0.3693502073176205,
      "epoch": 3.915099588294203,
      "grad_norm": 2.4375,
      "learning_rate": 1.7335555555555558e-05,
      "loss": 0.34,
      "mean_token_accuracy": 0.8858703788369894,
      "num_tokens": 58874383.0,
      "step": 2200
    },
    {
      "epoch": 3.915099588294203,
      "eval_chemistry_entropy": 0.40345276713371275,
      "eval_chemistry_loss": 0.7710844874382019,
      "eval_chemistry_mean_token_accuracy": 0.8157671246528625,
      "eval_chemistry_num_tokens": 58874383.0,
      "eval_chemistry_runtime": 34.5975,
      "eval_chemistry_samples_per_second": 14.452,
      "eval_chemistry_steps_per_second": 3.613,
      "step": 2200
    },
    {
      "epoch": 3.915099588294203,
      "eval_physics_entropy": 0.3868217145204544,
      "eval_physics_loss": 0.7909848690032959,
      "eval_physics_mean_token_accuracy": 0.8127978930473327,
      "eval_physics_num_tokens": 58874383.0,
      "eval_physics_runtime": 41.2517,
      "eval_physics_samples_per_second": 12.121,
      "eval_physics_steps_per_second": 3.03,
      "step": 2200
    },
    {
      "entropy": 0.3576618236489594,
      "epoch": 3.9329030822298874,
      "grad_norm": 2.09375,
      "learning_rate": 1.7313333333333336e-05,
      "loss": 0.3312,
      "mean_token_accuracy": 0.8888371895998717,
      "num_tokens": 59149972.0,
      "step": 2210
    },
    {
      "entropy": 0.3600709976628423,
      "epoch": 3.9507065761655724,
      "grad_norm": 2.140625,
      "learning_rate": 1.7291111111111113e-05,
      "loss": 0.3359,
      "mean_token_accuracy": 0.8871279675513506,
      "num_tokens": 59418591.0,
      "step": 2220
    },
    {
      "entropy": 0.36306885750964285,
      "epoch": 3.9685100701012574,
      "grad_norm": 2.484375,
      "learning_rate": 1.726888888888889e-05,
      "loss": 0.3344,
      "mean_token_accuracy": 0.8875064272433519,
      "num_tokens": 59691322.0,
      "step": 2230
    },
    {
      "entropy": 0.35846696645021436,
      "epoch": 3.9863135640369425,
      "grad_norm": 2.53125,
      "learning_rate": 1.724666666666667e-05,
      "loss": 0.3341,
      "mean_token_accuracy": 0.8886763751506805,
      "num_tokens": 59953838.0,
      "step": 2240
    },
    {
      "entropy": 0.34418878026546973,
      "epoch": 4.003560698787137,
      "grad_norm": 1.9453125,
      "learning_rate": 1.7224444444444446e-05,
      "loss": 0.3017,
      "mean_token_accuracy": 0.9004984659533347,
      "num_tokens": 60216049.0,
      "step": 2250
    },
    {
      "entropy": 0.2050149750430137,
      "epoch": 4.021364192722822,
      "grad_norm": 2.09375,
      "learning_rate": 1.7202222222222224e-05,
      "loss": 0.1646,
      "mean_token_accuracy": 0.9468532126396895,
      "num_tokens": 60490937.0,
      "step": 2260
    },
    {
      "entropy": 0.21089116097427904,
      "epoch": 4.039167686658507,
      "grad_norm": 2.265625,
      "learning_rate": 1.718e-05,
      "loss": 0.1616,
      "mean_token_accuracy": 0.9479435671120882,
      "num_tokens": 60755892.0,
      "step": 2270
    },
    {
      "entropy": 0.200906401546672,
      "epoch": 4.056971180594192,
      "grad_norm": 2.1875,
      "learning_rate": 1.715777777777778e-05,
      "loss": 0.1571,
      "mean_token_accuracy": 0.949157177656889,
      "num_tokens": 61030893.0,
      "step": 2280
    },
    {
      "entropy": 0.19398455056361855,
      "epoch": 4.074774674529876,
      "grad_norm": 2.046875,
      "learning_rate": 1.7135555555555557e-05,
      "loss": 0.1538,
      "mean_token_accuracy": 0.9495714385062456,
      "num_tokens": 61301458.0,
      "step": 2290
    },
    {
      "entropy": 0.20469776834361256,
      "epoch": 4.092578168465561,
      "grad_norm": 2.15625,
      "learning_rate": 1.7113333333333334e-05,
      "loss": 0.1598,
      "mean_token_accuracy": 0.9479261975735426,
      "num_tokens": 61574717.0,
      "step": 2300
    },
    {
      "epoch": 4.092578168465561,
      "eval_chemistry_entropy": 0.2952017867565155,
      "eval_chemistry_loss": 0.9488542079925537,
      "eval_chemistry_mean_token_accuracy": 0.8095356292724609,
      "eval_chemistry_num_tokens": 61574717.0,
      "eval_chemistry_runtime": 34.5891,
      "eval_chemistry_samples_per_second": 14.455,
      "eval_chemistry_steps_per_second": 3.614,
      "step": 2300
    },
    {
      "epoch": 4.092578168465561,
      "eval_physics_entropy": 0.2903540906906128,
      "eval_physics_loss": 0.966928243637085,
      "eval_physics_mean_token_accuracy": 0.8065616192817688,
      "eval_physics_num_tokens": 61574717.0,
      "eval_physics_runtime": 41.2483,
      "eval_physics_samples_per_second": 12.122,
      "eval_physics_steps_per_second": 3.03,
      "step": 2300
    },
    {
      "entropy": 0.2024691479280591,
      "epoch": 4.110381662401246,
      "grad_norm": 2.453125,
      "learning_rate": 1.7091111111111112e-05,
      "loss": 0.1633,
      "mean_token_accuracy": 0.947294469550252,
      "num_tokens": 61840186.0,
      "step": 2310
    },
    {
      "entropy": 0.20524093499407173,
      "epoch": 4.128185156336931,
      "grad_norm": 2.03125,
      "learning_rate": 1.706888888888889e-05,
      "loss": 0.1608,
      "mean_token_accuracy": 0.9480444736778736,
      "num_tokens": 62112844.0,
      "step": 2320
    },
    {
      "entropy": 0.20244754683226346,
      "epoch": 4.145988650272616,
      "grad_norm": 2.109375,
      "learning_rate": 1.704666666666667e-05,
      "loss": 0.158,
      "mean_token_accuracy": 0.9476622581481934,
      "num_tokens": 62382390.0,
      "step": 2330
    },
    {
      "entropy": 0.20337190260179341,
      "epoch": 4.163792144208301,
      "grad_norm": 2.09375,
      "learning_rate": 1.7024444444444445e-05,
      "loss": 0.1629,
      "mean_token_accuracy": 0.9467100698500872,
      "num_tokens": 62658026.0,
      "step": 2340
    },
    {
      "entropy": 0.20958826076239348,
      "epoch": 4.181595638143985,
      "grad_norm": 2.28125,
      "learning_rate": 1.7002222222222226e-05,
      "loss": 0.1678,
      "mean_token_accuracy": 0.9456365302205085,
      "num_tokens": 62916724.0,
      "step": 2350
    },
    {
      "entropy": 0.20147014455869794,
      "epoch": 4.19939913207967,
      "grad_norm": 2.203125,
      "learning_rate": 1.698e-05,
      "loss": 0.1598,
      "mean_token_accuracy": 0.9473498914390802,
      "num_tokens": 63195271.0,
      "step": 2360
    },
    {
      "entropy": 0.2052597001194954,
      "epoch": 4.2172026260153554,
      "grad_norm": 2.140625,
      "learning_rate": 1.695777777777778e-05,
      "loss": 0.1632,
      "mean_token_accuracy": 0.9463378667831421,
      "num_tokens": 63466066.0,
      "step": 2370
    },
    {
      "entropy": 0.20227183140814303,
      "epoch": 4.2350061199510405,
      "grad_norm": 2.296875,
      "learning_rate": 1.6935555555555555e-05,
      "loss": 0.1634,
      "mean_token_accuracy": 0.9463962618261575,
      "num_tokens": 63739678.0,
      "step": 2380
    },
    {
      "entropy": 0.20846639852970839,
      "epoch": 4.2528096138867255,
      "grad_norm": 2.25,
      "learning_rate": 1.6913333333333336e-05,
      "loss": 0.1656,
      "mean_token_accuracy": 0.9454980559647084,
      "num_tokens": 63999364.0,
      "step": 2390
    },
    {
      "entropy": 0.20496144848875703,
      "epoch": 4.2706131078224105,
      "grad_norm": 2.3125,
      "learning_rate": 1.689111111111111e-05,
      "loss": 0.1636,
      "mean_token_accuracy": 0.9468995247036218,
      "num_tokens": 64270501.0,
      "step": 2400
    },
    {
      "epoch": 4.2706131078224105,
      "eval_chemistry_entropy": 0.29572774827480314,
      "eval_chemistry_loss": 0.9564827680587769,
      "eval_chemistry_mean_token_accuracy": 0.8097796726226807,
      "eval_chemistry_num_tokens": 64270501.0,
      "eval_chemistry_runtime": 34.5923,
      "eval_chemistry_samples_per_second": 14.454,
      "eval_chemistry_steps_per_second": 3.614,
      "step": 2400
    },
    {
      "epoch": 4.2706131078224105,
      "eval_physics_entropy": 0.29065959858894347,
      "eval_physics_loss": 0.9759422540664673,
      "eval_physics_mean_token_accuracy": 0.8057192673683167,
      "eval_physics_num_tokens": 64270501.0,
      "eval_physics_runtime": 41.2353,
      "eval_physics_samples_per_second": 12.126,
      "eval_physics_steps_per_second": 3.031,
      "step": 2400
    },
    {
      "entropy": 0.20484122904017568,
      "epoch": 4.288416601758095,
      "grad_norm": 2.375,
      "learning_rate": 1.686888888888889e-05,
      "loss": 0.1653,
      "mean_token_accuracy": 0.9462751936167478,
      "num_tokens": 64533223.0,
      "step": 2410
    },
    {
      "entropy": 0.20944771389476954,
      "epoch": 4.30622009569378,
      "grad_norm": 2.375,
      "learning_rate": 1.684666666666667e-05,
      "loss": 0.1656,
      "mean_token_accuracy": 0.9454644002020359,
      "num_tokens": 64799113.0,
      "step": 2420
    },
    {
      "entropy": 0.20244625653140247,
      "epoch": 4.324023589629465,
      "grad_norm": 2.390625,
      "learning_rate": 1.6824444444444447e-05,
      "loss": 0.1647,
      "mean_token_accuracy": 0.945592139288783,
      "num_tokens": 65075459.0,
      "step": 2430
    },
    {
      "entropy": 0.20571904899552465,
      "epoch": 4.34182708356515,
      "grad_norm": 2.328125,
      "learning_rate": 1.6802222222222224e-05,
      "loss": 0.1663,
      "mean_token_accuracy": 0.9457980345934629,
      "num_tokens": 65341896.0,
      "step": 2440
    },
    {
      "entropy": 0.20644639227539302,
      "epoch": 4.359630577500835,
      "grad_norm": 2.125,
      "learning_rate": 1.6780000000000002e-05,
      "loss": 0.1662,
      "mean_token_accuracy": 0.9454721476882696,
      "num_tokens": 65614487.0,
      "step": 2450
    },
    {
      "entropy": 0.2118284550961107,
      "epoch": 4.37743407143652,
      "grad_norm": 2.359375,
      "learning_rate": 1.675777777777778e-05,
      "loss": 0.1719,
      "mean_token_accuracy": 0.9432908724993467,
      "num_tokens": 65887311.0,
      "step": 2460
    },
    {
      "entropy": 0.20808603088371455,
      "epoch": 4.395237565372204,
      "grad_norm": 2.25,
      "learning_rate": 1.6735555555555557e-05,
      "loss": 0.1666,
      "mean_token_accuracy": 0.9452798932790756,
      "num_tokens": 66151098.0,
      "step": 2470
    },
    {
      "entropy": 0.20479875253513455,
      "epoch": 4.413041059307889,
      "grad_norm": 2.59375,
      "learning_rate": 1.6713333333333335e-05,
      "loss": 0.167,
      "mean_token_accuracy": 0.9452142968773842,
      "num_tokens": 66417928.0,
      "step": 2480
    },
    {
      "entropy": 0.2097576475702226,
      "epoch": 4.430844553243574,
      "grad_norm": 2.53125,
      "learning_rate": 1.6691111111111112e-05,
      "loss": 0.1691,
      "mean_token_accuracy": 0.9440086893737316,
      "num_tokens": 66684481.0,
      "step": 2490
    },
    {
      "entropy": 0.21255099452100695,
      "epoch": 4.448648047179259,
      "grad_norm": 2.578125,
      "learning_rate": 1.666888888888889e-05,
      "loss": 0.1703,
      "mean_token_accuracy": 0.9436935927718878,
      "num_tokens": 66940215.0,
      "step": 2500
    },
    {
      "epoch": 4.448648047179259,
      "eval_chemistry_entropy": 0.2950603073835373,
      "eval_chemistry_loss": 0.9514251947402954,
      "eval_chemistry_mean_token_accuracy": 0.8093284454345703,
      "eval_chemistry_num_tokens": 66940215.0,
      "eval_chemistry_runtime": 34.5962,
      "eval_chemistry_samples_per_second": 14.452,
      "eval_chemistry_steps_per_second": 3.613,
      "step": 2500
    },
    {
      "epoch": 4.448648047179259,
      "eval_physics_entropy": 0.2886282374858856,
      "eval_physics_loss": 0.9700577259063721,
      "eval_physics_mean_token_accuracy": 0.807147349357605,
      "eval_physics_num_tokens": 66940215.0,
      "eval_physics_runtime": 40.9586,
      "eval_physics_samples_per_second": 12.207,
      "eval_physics_steps_per_second": 3.052,
      "step": 2500
    },
    {
      "entropy": 0.2105638356413692,
      "epoch": 4.466451541114944,
      "grad_norm": 2.34375,
      "learning_rate": 1.6646666666666668e-05,
      "loss": 0.172,
      "mean_token_accuracy": 0.9435432504862546,
      "num_tokens": 67207177.0,
      "step": 2510
    },
    {
      "entropy": 0.20986317647621036,
      "epoch": 4.484255035050628,
      "grad_norm": 2.25,
      "learning_rate": 1.6624444444444445e-05,
      "loss": 0.1691,
      "mean_token_accuracy": 0.9445763655006886,
      "num_tokens": 67477558.0,
      "step": 2520
    },
    {
      "entropy": 0.20789673756808041,
      "epoch": 4.502058528986313,
      "grad_norm": 2.578125,
      "learning_rate": 1.6602222222222223e-05,
      "loss": 0.1682,
      "mean_token_accuracy": 0.9449898913502693,
      "num_tokens": 67741434.0,
      "step": 2530
    },
    {
      "entropy": 0.21249219826422633,
      "epoch": 4.519862022921998,
      "grad_norm": 2.28125,
      "learning_rate": 1.658e-05,
      "loss": 0.1735,
      "mean_token_accuracy": 0.9431874167174101,
      "num_tokens": 68004497.0,
      "step": 2540
    },
    {
      "entropy": 0.2065403588116169,
      "epoch": 4.537665516857683,
      "grad_norm": 2.453125,
      "learning_rate": 1.6557777777777778e-05,
      "loss": 0.1687,
      "mean_token_accuracy": 0.9439776003360748,
      "num_tokens": 68276413.0,
      "step": 2550
    },
    {
      "entropy": 0.20407384247519075,
      "epoch": 4.555469010793368,
      "grad_norm": 2.484375,
      "learning_rate": 1.6535555555555556e-05,
      "loss": 0.1652,
      "mean_token_accuracy": 0.9454120360314846,
      "num_tokens": 68552989.0,
      "step": 2560
    },
    {
      "entropy": 0.21013034167699515,
      "epoch": 4.5732725047290534,
      "grad_norm": 2.390625,
      "learning_rate": 1.6513333333333333e-05,
      "loss": 0.1705,
      "mean_token_accuracy": 0.9439001563936472,
      "num_tokens": 68816878.0,
      "step": 2570
    },
    {
      "entropy": 0.20951900808140636,
      "epoch": 4.591075998664738,
      "grad_norm": 2.3125,
      "learning_rate": 1.6491111111111114e-05,
      "loss": 0.1705,
      "mean_token_accuracy": 0.94357103779912,
      "num_tokens": 69078503.0,
      "step": 2580
    },
    {
      "entropy": 0.20075970953330396,
      "epoch": 4.608879492600423,
      "grad_norm": 2.296875,
      "learning_rate": 1.646888888888889e-05,
      "loss": 0.1636,
      "mean_token_accuracy": 0.9454992815852166,
      "num_tokens": 69357964.0,
      "step": 2590
    },
    {
      "entropy": 0.2181381014175713,
      "epoch": 4.626682986536108,
      "grad_norm": 2.5,
      "learning_rate": 1.644666666666667e-05,
      "loss": 0.175,
      "mean_token_accuracy": 0.9416370660066604,
      "num_tokens": 69615568.0,
      "step": 2600
    },
    {
      "epoch": 4.626682986536108,
      "eval_chemistry_entropy": 0.29557589519023897,
      "eval_chemistry_loss": 0.9526376724243164,
      "eval_chemistry_mean_token_accuracy": 0.8096025767326355,
      "eval_chemistry_num_tokens": 69615568.0,
      "eval_chemistry_runtime": 34.6155,
      "eval_chemistry_samples_per_second": 14.444,
      "eval_chemistry_steps_per_second": 3.611,
      "step": 2600
    },
    {
      "epoch": 4.626682986536108,
      "eval_physics_entropy": 0.288856055855751,
      "eval_physics_loss": 0.975040078163147,
      "eval_physics_mean_token_accuracy": 0.8063538899421692,
      "eval_physics_num_tokens": 69615568.0,
      "eval_physics_runtime": 41.2472,
      "eval_physics_samples_per_second": 12.122,
      "eval_physics_steps_per_second": 3.031,
      "step": 2600
    },
    {
      "entropy": 0.2097658845130354,
      "epoch": 4.644486480471793,
      "grad_norm": 2.375,
      "learning_rate": 1.6424444444444444e-05,
      "loss": 0.1713,
      "mean_token_accuracy": 0.9439201038330793,
      "num_tokens": 69882933.0,
      "step": 2610
    },
    {
      "entropy": 0.21336092464625836,
      "epoch": 4.662289974407478,
      "grad_norm": 2.40625,
      "learning_rate": 1.6402222222222225e-05,
      "loss": 0.1755,
      "mean_token_accuracy": 0.9419367101043463,
      "num_tokens": 70147891.0,
      "step": 2620
    },
    {
      "entropy": 0.21264315508306025,
      "epoch": 4.680093468343163,
      "grad_norm": 2.546875,
      "learning_rate": 1.638e-05,
      "loss": 0.1717,
      "mean_token_accuracy": 0.9433247562497854,
      "num_tokens": 70411454.0,
      "step": 2630
    },
    {
      "entropy": 0.21375338733196259,
      "epoch": 4.697896962278847,
      "grad_norm": 2.421875,
      "learning_rate": 1.635777777777778e-05,
      "loss": 0.1748,
      "mean_token_accuracy": 0.9424599815160036,
      "num_tokens": 70673950.0,
      "step": 2640
    },
    {
      "entropy": 0.21311874822713434,
      "epoch": 4.715700456214532,
      "grad_norm": 2.40625,
      "learning_rate": 1.6335555555555558e-05,
      "loss": 0.1725,
      "mean_token_accuracy": 0.9424877449870109,
      "num_tokens": 70933220.0,
      "step": 2650
    },
    {
      "entropy": 0.20578455226495862,
      "epoch": 4.733503950150217,
      "grad_norm": 2.46875,
      "learning_rate": 1.6313333333333335e-05,
      "loss": 0.168,
      "mean_token_accuracy": 0.9445531789213419,
      "num_tokens": 71202429.0,
      "step": 2660
    },
    {
      "entropy": 0.2096489442512393,
      "epoch": 4.751307444085902,
      "grad_norm": 2.265625,
      "learning_rate": 1.6291111111111113e-05,
      "loss": 0.171,
      "mean_token_accuracy": 0.9438163749873638,
      "num_tokens": 71469322.0,
      "step": 2670
    },
    {
      "entropy": 0.21064688651822508,
      "epoch": 4.769110938021587,
      "grad_norm": 2.578125,
      "learning_rate": 1.626888888888889e-05,
      "loss": 0.1711,
      "mean_token_accuracy": 0.9425655394792557,
      "num_tokens": 71736522.0,
      "step": 2680
    },
    {
      "entropy": 0.21128194066695868,
      "epoch": 4.786914431957271,
      "grad_norm": 2.359375,
      "learning_rate": 1.6246666666666668e-05,
      "loss": 0.1732,
      "mean_token_accuracy": 0.9428361691534519,
      "num_tokens": 71999537.0,
      "step": 2690
    },
    {
      "entropy": 0.21140474304556847,
      "epoch": 4.804717925892956,
      "grad_norm": 2.265625,
      "learning_rate": 1.6224444444444446e-05,
      "loss": 0.1727,
      "mean_token_accuracy": 0.9425872072577477,
      "num_tokens": 72269735.0,
      "step": 2700
    },
    {
      "epoch": 4.804717925892956,
      "eval_chemistry_entropy": 0.2981076955795288,
      "eval_chemistry_loss": 0.9487899541854858,
      "eval_chemistry_mean_token_accuracy": 0.8097419013977051,
      "eval_chemistry_num_tokens": 72269735.0,
      "eval_chemistry_runtime": 34.6466,
      "eval_chemistry_samples_per_second": 14.431,
      "eval_chemistry_steps_per_second": 3.608,
      "step": 2700
    },
    {
      "epoch": 4.804717925892956,
      "eval_physics_entropy": 0.29301643872261046,
      "eval_physics_loss": 0.9743095636367798,
      "eval_physics_mean_token_accuracy": 0.8055673995018006,
      "eval_physics_num_tokens": 72269735.0,
      "eval_physics_runtime": 41.2856,
      "eval_physics_samples_per_second": 12.111,
      "eval_physics_steps_per_second": 3.028,
      "step": 2700
    },
    {
      "entropy": 0.20992932757362723,
      "epoch": 4.822521419828641,
      "grad_norm": 2.3125,
      "learning_rate": 1.6202222222222223e-05,
      "loss": 0.1695,
      "mean_token_accuracy": 0.9435645472258329,
      "num_tokens": 72537744.0,
      "step": 2710
    },
    {
      "entropy": 0.21736142109148204,
      "epoch": 4.840324913764326,
      "grad_norm": 2.40625,
      "learning_rate": 1.618e-05,
      "loss": 0.1749,
      "mean_token_accuracy": 0.9426166765391827,
      "num_tokens": 72801656.0,
      "step": 2720
    },
    {
      "entropy": 0.20337292733602225,
      "epoch": 4.858128407700011,
      "grad_norm": 2.34375,
      "learning_rate": 1.615777777777778e-05,
      "loss": 0.166,
      "mean_token_accuracy": 0.9448446430265903,
      "num_tokens": 73075773.0,
      "step": 2730
    },
    {
      "entropy": 0.20302016455680133,
      "epoch": 4.875931901635696,
      "grad_norm": 2.265625,
      "learning_rate": 1.6135555555555556e-05,
      "loss": 0.1673,
      "mean_token_accuracy": 0.9446327567100525,
      "num_tokens": 73353692.0,
      "step": 2740
    },
    {
      "entropy": 0.2136679151561111,
      "epoch": 4.893735395571381,
      "grad_norm": 2.546875,
      "learning_rate": 1.6113333333333334e-05,
      "loss": 0.1753,
      "mean_token_accuracy": 0.9421532608568668,
      "num_tokens": 73620577.0,
      "step": 2750
    },
    {
      "entropy": 0.2027943292632699,
      "epoch": 4.9115388895070655,
      "grad_norm": 2.234375,
      "learning_rate": 1.609111111111111e-05,
      "loss": 0.166,
      "mean_token_accuracy": 0.945094608142972,
      "num_tokens": 73888689.0,
      "step": 2760
    },
    {
      "entropy": 0.21415654346346855,
      "epoch": 4.9293423834427506,
      "grad_norm": 2.4375,
      "learning_rate": 1.606888888888889e-05,
      "loss": 0.1744,
      "mean_token_accuracy": 0.942426010593772,
      "num_tokens": 74150672.0,
      "step": 2770
    },
    {
      "entropy": 0.21133902384899556,
      "epoch": 4.947145877378436,
      "grad_norm": 2.25,
      "learning_rate": 1.6046666666666667e-05,
      "loss": 0.1728,
      "mean_token_accuracy": 0.9424780901521445,
      "num_tokens": 74419554.0,
      "step": 2780
    },
    {
      "entropy": 0.21058966973796486,
      "epoch": 4.964949371314121,
      "grad_norm": 2.34375,
      "learning_rate": 1.6024444444444444e-05,
      "loss": 0.1741,
      "mean_token_accuracy": 0.9428736094385386,
      "num_tokens": 74686704.0,
      "step": 2790
    },
    {
      "entropy": 0.21359067875891924,
      "epoch": 4.982752865249806,
      "grad_norm": 2.59375,
      "learning_rate": 1.6002222222222222e-05,
      "loss": 0.1744,
      "mean_token_accuracy": 0.942025650665164,
      "num_tokens": 74947391.0,
      "step": 2800
    },
    {
      "epoch": 4.982752865249806,
      "eval_chemistry_entropy": 0.299175696849823,
      "eval_chemistry_loss": 0.9473890662193298,
      "eval_chemistry_mean_token_accuracy": 0.8102310700416565,
      "eval_chemistry_num_tokens": 74947391.0,
      "eval_chemistry_runtime": 34.9666,
      "eval_chemistry_samples_per_second": 14.299,
      "eval_chemistry_steps_per_second": 3.575,
      "step": 2800
    },
    {
      "epoch": 4.982752865249806,
      "eval_physics_entropy": 0.29367818546295166,
      "eval_physics_loss": 0.9738690257072449,
      "eval_physics_mean_token_accuracy": 0.8059637842178344,
      "eval_physics_num_tokens": 74947391.0,
      "eval_physics_runtime": 41.2321,
      "eval_physics_samples_per_second": 12.126,
      "eval_physics_steps_per_second": 3.032,
      "step": 2800
    },
    {
      "entropy": 0.21012076139450073,
      "epoch": 5.0,
      "grad_norm": 2.96875,
      "learning_rate": 1.5980000000000003e-05,
      "loss": 0.1708,
      "mean_token_accuracy": 0.9433038357765444,
      "num_tokens": 75199650.0,
      "step": 2810
    },
    {
      "entropy": 0.14696180289611221,
      "epoch": 5.017803493935685,
      "grad_norm": 1.78125,
      "learning_rate": 1.5957777777777777e-05,
      "loss": 0.0849,
      "mean_token_accuracy": 0.9769088149070739,
      "num_tokens": 75457508.0,
      "step": 2820
    },
    {
      "entropy": 0.1315455799922347,
      "epoch": 5.03560698787137,
      "grad_norm": 1.5234375,
      "learning_rate": 1.5935555555555558e-05,
      "loss": 0.0803,
      "mean_token_accuracy": 0.9773992955684662,
      "num_tokens": 75731578.0,
      "step": 2830
    },
    {
      "entropy": 0.12918212548829616,
      "epoch": 5.053410481807054,
      "grad_norm": 1.6484375,
      "learning_rate": 1.5913333333333332e-05,
      "loss": 0.0793,
      "mean_token_accuracy": 0.9779823809862137,
      "num_tokens": 76008875.0,
      "step": 2840
    },
    {
      "entropy": 0.1309114085044712,
      "epoch": 5.071213975742739,
      "grad_norm": 1.625,
      "learning_rate": 1.5891111111111113e-05,
      "loss": 0.0811,
      "mean_token_accuracy": 0.9771320998668671,
      "num_tokens": 76281818.0,
      "step": 2850
    },
    {
      "entropy": 0.1301674212096259,
      "epoch": 5.089017469678424,
      "grad_norm": 1.5234375,
      "learning_rate": 1.5868888888888888e-05,
      "loss": 0.0811,
      "mean_token_accuracy": 0.976826249063015,
      "num_tokens": 76549100.0,
      "step": 2860
    },
    {
      "entropy": 0.13395183440297842,
      "epoch": 5.106820963614109,
      "grad_norm": 1.78125,
      "learning_rate": 1.584666666666667e-05,
      "loss": 0.0838,
      "mean_token_accuracy": 0.9762667279690505,
      "num_tokens": 76815070.0,
      "step": 2870
    },
    {
      "entropy": 0.1291714665479958,
      "epoch": 5.124624457549794,
      "grad_norm": 1.78125,
      "learning_rate": 1.5824444444444446e-05,
      "loss": 0.0818,
      "mean_token_accuracy": 0.9767425782978535,
      "num_tokens": 77090788.0,
      "step": 2880
    },
    {
      "entropy": 0.13371132141910494,
      "epoch": 5.142427951485479,
      "grad_norm": 1.7109375,
      "learning_rate": 1.5802222222222224e-05,
      "loss": 0.0852,
      "mean_token_accuracy": 0.9760543793439865,
      "num_tokens": 77356917.0,
      "step": 2890
    },
    {
      "entropy": 0.13178068292327225,
      "epoch": 5.1602314454211635,
      "grad_norm": 1.7265625,
      "learning_rate": 1.578e-05,
      "loss": 0.0844,
      "mean_token_accuracy": 0.9758262522518635,
      "num_tokens": 77619491.0,
      "step": 2900
    },
    {
      "epoch": 5.1602314454211635,
      "eval_chemistry_entropy": 0.25860024750232696,
      "eval_chemistry_loss": 1.0449376106262207,
      "eval_chemistry_mean_token_accuracy": 0.8088349919319153,
      "eval_chemistry_num_tokens": 77619491.0,
      "eval_chemistry_runtime": 34.6148,
      "eval_chemistry_samples_per_second": 14.445,
      "eval_chemistry_steps_per_second": 3.611,
      "step": 2900
    },
    {
      "epoch": 5.1602314454211635,
      "eval_physics_entropy": 0.253796443104744,
      "eval_physics_loss": 1.0744327306747437,
      "eval_physics_mean_token_accuracy": 0.8047778859138489,
      "eval_physics_num_tokens": 77619491.0,
      "eval_physics_runtime": 41.2839,
      "eval_physics_samples_per_second": 12.111,
      "eval_physics_steps_per_second": 3.028,
      "step": 2900
    },
    {
      "entropy": 0.13181702424772085,
      "epoch": 5.1780349393568486,
      "grad_norm": 1.578125,
      "learning_rate": 1.575777777777778e-05,
      "loss": 0.0843,
      "mean_token_accuracy": 0.9760214772075415,
      "num_tokens": 77885239.0,
      "step": 2910
    },
    {
      "entropy": 0.1320621079299599,
      "epoch": 5.195838433292534,
      "grad_norm": 1.7734375,
      "learning_rate": 1.5735555555555557e-05,
      "loss": 0.0845,
      "mean_token_accuracy": 0.9757996380329133,
      "num_tokens": 78154475.0,
      "step": 2920
    },
    {
      "entropy": 0.13251891094259918,
      "epoch": 5.213641927228219,
      "grad_norm": 1.9609375,
      "learning_rate": 1.5713333333333334e-05,
      "loss": 0.0859,
      "mean_token_accuracy": 0.975741833820939,
      "num_tokens": 78418518.0,
      "step": 2930
    },
    {
      "entropy": 0.1304515670053661,
      "epoch": 5.231445421163904,
      "grad_norm": 1.7265625,
      "learning_rate": 1.5691111111111112e-05,
      "loss": 0.0856,
      "mean_token_accuracy": 0.9751247983425856,
      "num_tokens": 78687867.0,
      "step": 2940
    },
    {
      "entropy": 0.13110208376310767,
      "epoch": 5.249248915099589,
      "grad_norm": 1.6640625,
      "learning_rate": 1.5668888888888893e-05,
      "loss": 0.0861,
      "mean_token_accuracy": 0.9753684744238853,
      "num_tokens": 78954512.0,
      "step": 2950
    },
    {
      "entropy": 0.13201983934268355,
      "epoch": 5.267052409035273,
      "grad_norm": 1.765625,
      "learning_rate": 1.5646666666666667e-05,
      "loss": 0.0852,
      "mean_token_accuracy": 0.9757239740341902,
      "num_tokens": 79219529.0,
      "step": 2960
    },
    {
      "entropy": 0.13015602058731018,
      "epoch": 5.284855902970958,
      "grad_norm": 1.671875,
      "learning_rate": 1.5624444444444448e-05,
      "loss": 0.0831,
      "mean_token_accuracy": 0.9763064742088318,
      "num_tokens": 79494122.0,
      "step": 2970
    },
    {
      "entropy": 0.12984587037935852,
      "epoch": 5.302659396906643,
      "grad_norm": 1.734375,
      "learning_rate": 1.5602222222222222e-05,
      "loss": 0.0844,
      "mean_token_accuracy": 0.9757621839642525,
      "num_tokens": 79766616.0,
      "step": 2980
    },
    {
      "entropy": 0.13055733423680066,
      "epoch": 5.320462890842328,
      "grad_norm": 1.828125,
      "learning_rate": 1.5580000000000003e-05,
      "loss": 0.0852,
      "mean_token_accuracy": 0.9758519910275936,
      "num_tokens": 80028355.0,
      "step": 2990
    },
    {
      "entropy": 0.1292526103556156,
      "epoch": 5.338266384778013,
      "grad_norm": 1.78125,
      "learning_rate": 1.5557777777777778e-05,
      "loss": 0.0849,
      "mean_token_accuracy": 0.9756632026284933,
      "num_tokens": 80296531.0,
      "step": 3000
    },
    {
      "epoch": 5.338266384778013,
      "eval_chemistry_entropy": 0.2542178691625595,
      "eval_chemistry_loss": 1.0537911653518677,
      "eval_chemistry_mean_token_accuracy": 0.8096120843887329,
      "eval_chemistry_num_tokens": 80296531.0,
      "eval_chemistry_runtime": 34.603,
      "eval_chemistry_samples_per_second": 14.45,
      "eval_chemistry_steps_per_second": 3.612,
      "step": 3000
    },
    {
      "epoch": 5.338266384778013,
      "eval_physics_entropy": 0.25119746136665344,
      "eval_physics_loss": 1.0833330154418945,
      "eval_physics_mean_token_accuracy": 0.8041757836341857,
      "eval_physics_num_tokens": 80296531.0,
      "eval_physics_runtime": 40.9933,
      "eval_physics_samples_per_second": 12.197,
      "eval_physics_steps_per_second": 3.049,
      "step": 3000
    },
    {
      "entropy": 0.1336166656576097,
      "epoch": 5.356069878713697,
      "grad_norm": 1.7734375,
      "learning_rate": 1.553555555555556e-05,
      "loss": 0.0872,
      "mean_token_accuracy": 0.9749985877424479,
      "num_tokens": 80556132.0,
      "step": 3010
    },
    {
      "entropy": 0.13146277102641762,
      "epoch": 5.373873372649382,
      "grad_norm": 1.7421875,
      "learning_rate": 1.5513333333333333e-05,
      "loss": 0.0855,
      "mean_token_accuracy": 0.9750809911638498,
      "num_tokens": 80823194.0,
      "step": 3020
    },
    {
      "entropy": 0.13275023545138537,
      "epoch": 5.391676866585067,
      "grad_norm": 1.7734375,
      "learning_rate": 1.5491111111111114e-05,
      "loss": 0.0859,
      "mean_token_accuracy": 0.9754689898341894,
      "num_tokens": 81092927.0,
      "step": 3030
    },
    {
      "entropy": 0.13171642664819955,
      "epoch": 5.409480360520752,
      "grad_norm": 1.53125,
      "learning_rate": 1.546888888888889e-05,
      "loss": 0.0849,
      "mean_token_accuracy": 0.9756545100361109,
      "num_tokens": 81365817.0,
      "step": 3040
    },
    {
      "entropy": 0.1337125868536532,
      "epoch": 5.427283854456437,
      "grad_norm": 1.7890625,
      "learning_rate": 1.544666666666667e-05,
      "loss": 0.0885,
      "mean_token_accuracy": 0.9743953909724951,
      "num_tokens": 81634089.0,
      "step": 3050
    },
    {
      "entropy": 0.13215969274751843,
      "epoch": 5.445087348392122,
      "grad_norm": 1.8046875,
      "learning_rate": 1.5424444444444447e-05,
      "loss": 0.0861,
      "mean_token_accuracy": 0.9752333201467991,
      "num_tokens": 81896949.0,
      "step": 3060
    },
    {
      "entropy": 0.13062844248488545,
      "epoch": 5.462890842327806,
      "grad_norm": 1.640625,
      "learning_rate": 1.5402222222222224e-05,
      "loss": 0.0862,
      "mean_token_accuracy": 0.9753217443823814,
      "num_tokens": 82163467.0,
      "step": 3070
    },
    {
      "entropy": 0.12999712373130023,
      "epoch": 5.4806943362634915,
      "grad_norm": 1.8203125,
      "learning_rate": 1.5380000000000002e-05,
      "loss": 0.0856,
      "mean_token_accuracy": 0.975192042067647,
      "num_tokens": 82438582.0,
      "step": 3080
    },
    {
      "entropy": 0.13370983600616454,
      "epoch": 5.4984978301991765,
      "grad_norm": 1.65625,
      "learning_rate": 1.535777777777778e-05,
      "loss": 0.0858,
      "mean_token_accuracy": 0.975127911195159,
      "num_tokens": 82709044.0,
      "step": 3090
    },
    {
      "entropy": 0.13575179101899265,
      "epoch": 5.5163013241348615,
      "grad_norm": 1.8203125,
      "learning_rate": 1.5335555555555557e-05,
      "loss": 0.0879,
      "mean_token_accuracy": 0.9744067292660474,
      "num_tokens": 82971488.0,
      "step": 3100
    },
    {
      "epoch": 5.5163013241348615,
      "eval_chemistry_entropy": 0.25467691576480866,
      "eval_chemistry_loss": 1.057073712348938,
      "eval_chemistry_mean_token_accuracy": 0.8083673243522644,
      "eval_chemistry_num_tokens": 82971488.0,
      "eval_chemistry_runtime": 34.6096,
      "eval_chemistry_samples_per_second": 14.447,
      "eval_chemistry_steps_per_second": 3.612,
      "step": 3100
    },
    {
      "epoch": 5.5163013241348615,
      "eval_physics_entropy": 0.25039119863510134,
      "eval_physics_loss": 1.0901731252670288,
      "eval_physics_mean_token_accuracy": 0.8044232959747315,
      "eval_physics_num_tokens": 82971488.0,
      "eval_physics_runtime": 41.2719,
      "eval_physics_samples_per_second": 12.115,
      "eval_physics_steps_per_second": 3.029,
      "step": 3100
    },
    {
      "entropy": 0.13085725898854433,
      "epoch": 5.5341048180705465,
      "grad_norm": 1.8359375,
      "learning_rate": 1.5313333333333335e-05,
      "loss": 0.0853,
      "mean_token_accuracy": 0.9751925073564053,
      "num_tokens": 83243796.0,
      "step": 3110
    },
    {
      "entropy": 0.13076143795624376,
      "epoch": 5.551908312006232,
      "grad_norm": 1.65625,
      "learning_rate": 1.5291111111111112e-05,
      "loss": 0.0863,
      "mean_token_accuracy": 0.974893382564187,
      "num_tokens": 83525241.0,
      "step": 3120
    },
    {
      "entropy": 0.13085856032557786,
      "epoch": 5.569711805941916,
      "grad_norm": 1.5859375,
      "learning_rate": 1.526888888888889e-05,
      "loss": 0.0866,
      "mean_token_accuracy": 0.9745663721114397,
      "num_tokens": 83794521.0,
      "step": 3130
    },
    {
      "entropy": 0.13409928008913993,
      "epoch": 5.587515299877601,
      "grad_norm": 1.8125,
      "learning_rate": 1.5246666666666668e-05,
      "loss": 0.0885,
      "mean_token_accuracy": 0.9742066409438849,
      "num_tokens": 84053712.0,
      "step": 3140
    },
    {
      "entropy": 0.13780566449277104,
      "epoch": 5.605318793813286,
      "grad_norm": 1.8671875,
      "learning_rate": 1.5224444444444447e-05,
      "loss": 0.0913,
      "mean_token_accuracy": 0.9734075956046582,
      "num_tokens": 84314508.0,
      "step": 3150
    },
    {
      "entropy": 0.1349004504736513,
      "epoch": 5.623122287748971,
      "grad_norm": 1.875,
      "learning_rate": 1.5202222222222223e-05,
      "loss": 0.0885,
      "mean_token_accuracy": 0.9740460023283959,
      "num_tokens": 84585078.0,
      "step": 3160
    },
    {
      "entropy": 0.1312039703130722,
      "epoch": 5.640925781684656,
      "grad_norm": 1.8125,
      "learning_rate": 1.5180000000000002e-05,
      "loss": 0.0864,
      "mean_token_accuracy": 0.9751023899763822,
      "num_tokens": 84850440.0,
      "step": 3170
    },
    {
      "entropy": 0.13376746168360115,
      "epoch": 5.65872927562034,
      "grad_norm": 1.84375,
      "learning_rate": 1.5157777777777778e-05,
      "loss": 0.0884,
      "mean_token_accuracy": 0.9744480658322573,
      "num_tokens": 85108727.0,
      "step": 3180
    },
    {
      "entropy": 0.1314602940576151,
      "epoch": 5.676532769556025,
      "grad_norm": 1.9375,
      "learning_rate": 1.5135555555555557e-05,
      "loss": 0.0876,
      "mean_token_accuracy": 0.9745025612413883,
      "num_tokens": 85375410.0,
      "step": 3190
    },
    {
      "entropy": 0.13288075104355812,
      "epoch": 5.69433626349171,
      "grad_norm": 1.734375,
      "learning_rate": 1.5113333333333335e-05,
      "loss": 0.0886,
      "mean_token_accuracy": 0.9740624383091927,
      "num_tokens": 85644389.0,
      "step": 3200
    },
    {
      "epoch": 5.69433626349171,
      "eval_chemistry_entropy": 0.25195063865184786,
      "eval_chemistry_loss": 1.0551501512527466,
      "eval_chemistry_mean_token_accuracy": 0.8095473246574402,
      "eval_chemistry_num_tokens": 85644389.0,
      "eval_chemistry_runtime": 34.6028,
      "eval_chemistry_samples_per_second": 14.45,
      "eval_chemistry_steps_per_second": 3.612,
      "step": 3200
    },
    {
      "epoch": 5.69433626349171,
      "eval_physics_entropy": 0.24716808009147645,
      "eval_physics_loss": 1.0878981351852417,
      "eval_physics_mean_token_accuracy": 0.8046164889335632,
      "eval_physics_num_tokens": 85644389.0,
      "eval_physics_runtime": 41.2548,
      "eval_physics_samples_per_second": 12.12,
      "eval_physics_steps_per_second": 3.03,
      "step": 3200
    },
    {
      "entropy": 0.13199728620238602,
      "epoch": 5.712139757427395,
      "grad_norm": 1.84375,
      "learning_rate": 1.5091111111111113e-05,
      "loss": 0.0863,
      "mean_token_accuracy": 0.9750527568161488,
      "num_tokens": 85913924.0,
      "step": 3210
    },
    {
      "entropy": 0.1298374376259744,
      "epoch": 5.72994325136308,
      "grad_norm": 1.75,
      "learning_rate": 1.506888888888889e-05,
      "loss": 0.0868,
      "mean_token_accuracy": 0.9743161331862211,
      "num_tokens": 86186499.0,
      "step": 3220
    },
    {
      "entropy": 0.13264909987337886,
      "epoch": 5.747746745298765,
      "grad_norm": 1.7109375,
      "learning_rate": 1.5046666666666668e-05,
      "loss": 0.0875,
      "mean_token_accuracy": 0.9744381692260504,
      "num_tokens": 86457529.0,
      "step": 3230
    },
    {
      "entropy": 0.13380379271693527,
      "epoch": 5.76555023923445,
      "grad_norm": 1.859375,
      "learning_rate": 1.5024444444444445e-05,
      "loss": 0.0895,
      "mean_token_accuracy": 0.9738269798457623,
      "num_tokens": 86721227.0,
      "step": 3240
    },
    {
      "entropy": 0.13263347847387194,
      "epoch": 5.783353733170134,
      "grad_norm": 2.0625,
      "learning_rate": 1.5002222222222223e-05,
      "loss": 0.0882,
      "mean_token_accuracy": 0.9740435156971217,
      "num_tokens": 86987700.0,
      "step": 3250
    },
    {
      "entropy": 0.13063039840199053,
      "epoch": 5.801157227105819,
      "grad_norm": 1.8359375,
      "learning_rate": 1.498e-05,
      "loss": 0.0874,
      "mean_token_accuracy": 0.9743896011263132,
      "num_tokens": 87253894.0,
      "step": 3260
    },
    {
      "entropy": 0.1305804959964007,
      "epoch": 5.818960721041504,
      "grad_norm": 1.734375,
      "learning_rate": 1.495777777777778e-05,
      "loss": 0.0872,
      "mean_token_accuracy": 0.9744400773197412,
      "num_tokens": 87524480.0,
      "step": 3270
    },
    {
      "entropy": 0.13412257907912134,
      "epoch": 5.8367642149771894,
      "grad_norm": 1.96875,
      "learning_rate": 1.4935555555555556e-05,
      "loss": 0.088,
      "mean_token_accuracy": 0.9747412655502558,
      "num_tokens": 87787503.0,
      "step": 3280
    },
    {
      "entropy": 0.13367903619073332,
      "epoch": 5.8545677089128745,
      "grad_norm": 1.875,
      "learning_rate": 1.4913333333333335e-05,
      "loss": 0.0887,
      "mean_token_accuracy": 0.97428372874856,
      "num_tokens": 88050314.0,
      "step": 3290
    },
    {
      "entropy": 0.13025642042048274,
      "epoch": 5.872371202848559,
      "grad_norm": 1.6796875,
      "learning_rate": 1.4891111111111111e-05,
      "loss": 0.0871,
      "mean_token_accuracy": 0.9744457863271236,
      "num_tokens": 88319767.0,
      "step": 3300
    },
    {
      "epoch": 5.872371202848559,
      "eval_chemistry_entropy": 0.2513621771335602,
      "eval_chemistry_loss": 1.0590453147888184,
      "eval_chemistry_mean_token_accuracy": 0.8089683670997619,
      "eval_chemistry_num_tokens": 88319767.0,
      "eval_chemistry_runtime": 34.6404,
      "eval_chemistry_samples_per_second": 14.434,
      "eval_chemistry_steps_per_second": 3.609,
      "step": 3300
    },
    {
      "epoch": 5.872371202848559,
      "eval_physics_entropy": 0.24686662149429323,
      "eval_physics_loss": 1.0921601057052612,
      "eval_physics_mean_token_accuracy": 0.8040507826805114,
      "eval_physics_num_tokens": 88319767.0,
      "eval_physics_runtime": 41.317,
      "eval_physics_samples_per_second": 12.102,
      "eval_physics_steps_per_second": 3.025,
      "step": 3300
    },
    {
      "entropy": 0.13620419343933463,
      "epoch": 5.890174696784244,
      "grad_norm": 1.8359375,
      "learning_rate": 1.486888888888889e-05,
      "loss": 0.0901,
      "mean_token_accuracy": 0.973667586222291,
      "num_tokens": 88584663.0,
      "step": 3310
    },
    {
      "entropy": 0.1310937553178519,
      "epoch": 5.907978190719929,
      "grad_norm": 1.8203125,
      "learning_rate": 1.4846666666666666e-05,
      "loss": 0.0873,
      "mean_token_accuracy": 0.9745351355522871,
      "num_tokens": 88853779.0,
      "step": 3320
    },
    {
      "entropy": 0.132945366948843,
      "epoch": 5.925781684655614,
      "grad_norm": 1.84375,
      "learning_rate": 1.4824444444444446e-05,
      "loss": 0.0884,
      "mean_token_accuracy": 0.9745597939938306,
      "num_tokens": 89119463.0,
      "step": 3330
    },
    {
      "entropy": 0.12810182510875165,
      "epoch": 5.943585178591299,
      "grad_norm": 1.7890625,
      "learning_rate": 1.4802222222222225e-05,
      "loss": 0.0859,
      "mean_token_accuracy": 0.9745643649250268,
      "num_tokens": 89395043.0,
      "step": 3340
    },
    {
      "entropy": 0.13572987951338292,
      "epoch": 5.961388672526984,
      "grad_norm": 1.828125,
      "learning_rate": 1.478e-05,
      "loss": 0.0893,
      "mean_token_accuracy": 0.9741614494472742,
      "num_tokens": 89659996.0,
      "step": 3350
    },
    {
      "entropy": 0.12838005665689706,
      "epoch": 5.979192166462668,
      "grad_norm": 1.96875,
      "learning_rate": 1.475777777777778e-05,
      "loss": 0.0859,
      "mean_token_accuracy": 0.9746335234493018,
      "num_tokens": 89924565.0,
      "step": 3360
    },
    {
      "entropy": 0.133992342883721,
      "epoch": 5.996995660398353,
      "grad_norm": 1.828125,
      "learning_rate": 1.4735555555555556e-05,
      "loss": 0.0882,
      "mean_token_accuracy": 0.9745088398456574,
      "num_tokens": 90191795.0,
      "step": 3370
    },
    {
      "entropy": 0.1073445541724082,
      "epoch": 6.014242795148548,
      "grad_norm": 1.0546875,
      "learning_rate": 1.4713333333333335e-05,
      "loss": 0.062,
      "mean_token_accuracy": 0.9837527901895584,
      "num_tokens": 90445437.0,
      "step": 3380
    },
    {
      "entropy": 0.08888966594822705,
      "epoch": 6.032046289084232,
      "grad_norm": 1.0546875,
      "learning_rate": 1.4691111111111111e-05,
      "loss": 0.0525,
      "mean_token_accuracy": 0.9863325536251069,
      "num_tokens": 90720244.0,
      "step": 3390
    },
    {
      "entropy": 0.09145279829390347,
      "epoch": 6.049849783019917,
      "grad_norm": 1.0546875,
      "learning_rate": 1.466888888888889e-05,
      "loss": 0.0527,
      "mean_token_accuracy": 0.9857543986290693,
      "num_tokens": 90988706.0,
      "step": 3400
    },
    {
      "epoch": 6.049849783019917,
      "eval_chemistry_entropy": 0.23236147463321685,
      "eval_chemistry_loss": 1.1179053783416748,
      "eval_chemistry_mean_token_accuracy": 0.8094306111335754,
      "eval_chemistry_num_tokens": 90988706.0,
      "eval_chemistry_runtime": 34.5943,
      "eval_chemistry_samples_per_second": 14.453,
      "eval_chemistry_steps_per_second": 3.613,
      "step": 3400
    },
    {
      "epoch": 6.049849783019917,
      "eval_physics_entropy": 0.22796263718605042,
      "eval_physics_loss": 1.1574972867965698,
      "eval_physics_mean_token_accuracy": 0.8041394581794739,
      "eval_physics_num_tokens": 90988706.0,
      "eval_physics_runtime": 41.2326,
      "eval_physics_samples_per_second": 12.126,
      "eval_physics_steps_per_second": 3.032,
      "step": 3400
    },
    {
      "entropy": 0.09066825113259255,
      "epoch": 6.067653276955602,
      "grad_norm": 1.125,
      "learning_rate": 1.4646666666666666e-05,
      "loss": 0.053,
      "mean_token_accuracy": 0.9855694826692343,
      "num_tokens": 91255995.0,
      "step": 3410
    },
    {
      "entropy": 0.09201736019458621,
      "epoch": 6.0854567708912874,
      "grad_norm": 1.328125,
      "learning_rate": 1.4624444444444446e-05,
      "loss": 0.0543,
      "mean_token_accuracy": 0.9853279072791338,
      "num_tokens": 91524803.0,
      "step": 3420
    },
    {
      "entropy": 0.0917032704455778,
      "epoch": 6.1032602648269725,
      "grad_norm": 1.359375,
      "learning_rate": 1.4602222222222225e-05,
      "loss": 0.0545,
      "mean_token_accuracy": 0.9851662259548902,
      "num_tokens": 91788457.0,
      "step": 3430
    },
    {
      "entropy": 0.08859047300647944,
      "epoch": 6.1210637587626575,
      "grad_norm": 1.15625,
      "learning_rate": 1.4580000000000001e-05,
      "loss": 0.0531,
      "mean_token_accuracy": 0.9854934360831976,
      "num_tokens": 92061062.0,
      "step": 3440
    },
    {
      "entropy": 0.0946397572522983,
      "epoch": 6.138867252698342,
      "grad_norm": 1.078125,
      "learning_rate": 1.455777777777778e-05,
      "loss": 0.0557,
      "mean_token_accuracy": 0.9848660323768854,
      "num_tokens": 92323734.0,
      "step": 3450
    },
    {
      "entropy": 0.09210386867634952,
      "epoch": 6.156670746634027,
      "grad_norm": 1.2578125,
      "learning_rate": 1.4535555555555556e-05,
      "loss": 0.0548,
      "mean_token_accuracy": 0.9850046575069428,
      "num_tokens": 92588818.0,
      "step": 3460
    },
    {
      "entropy": 0.08919113818556071,
      "epoch": 6.174474240569712,
      "grad_norm": 1.1640625,
      "learning_rate": 1.4513333333333336e-05,
      "loss": 0.053,
      "mean_token_accuracy": 0.98563387170434,
      "num_tokens": 92863979.0,
      "step": 3470
    },
    {
      "entropy": 0.09293870662804693,
      "epoch": 6.192277734505397,
      "grad_norm": 1.390625,
      "learning_rate": 1.4491111111111111e-05,
      "loss": 0.0556,
      "mean_token_accuracy": 0.9848345395177602,
      "num_tokens": 93125523.0,
      "step": 3480
    },
    {
      "entropy": 0.09469078800175339,
      "epoch": 6.210081228441082,
      "grad_norm": 1.1953125,
      "learning_rate": 1.446888888888889e-05,
      "loss": 0.057,
      "mean_token_accuracy": 0.9842254180461169,
      "num_tokens": 93388128.0,
      "step": 3490
    },
    {
      "entropy": 0.09291742001660168,
      "epoch": 6.227884722376767,
      "grad_norm": 1.25,
      "learning_rate": 1.4446666666666668e-05,
      "loss": 0.0559,
      "mean_token_accuracy": 0.9847728945314884,
      "num_tokens": 93656779.0,
      "step": 3500
    },
    {
      "epoch": 6.227884722376767,
      "eval_chemistry_entropy": 0.23241128754615784,
      "eval_chemistry_loss": 1.1239092350006104,
      "eval_chemistry_mean_token_accuracy": 0.8096166286468506,
      "eval_chemistry_num_tokens": 93656779.0,
      "eval_chemistry_runtime": 34.6187,
      "eval_chemistry_samples_per_second": 14.443,
      "eval_chemistry_steps_per_second": 3.611,
      "step": 3500
    },
    {
      "epoch": 6.227884722376767,
      "eval_physics_entropy": 0.22787879633903504,
      "eval_physics_loss": 1.1623262166976929,
      "eval_physics_mean_token_accuracy": 0.8041429243087769,
      "eval_physics_num_tokens": 93656779.0,
      "eval_physics_runtime": 41.3044,
      "eval_physics_samples_per_second": 12.105,
      "eval_physics_steps_per_second": 3.026,
      "step": 3500
    },
    {
      "entropy": 0.09307919752318412,
      "epoch": 6.245688216312451,
      "grad_norm": 1.28125,
      "learning_rate": 1.4424444444444446e-05,
      "loss": 0.0555,
      "mean_token_accuracy": 0.9850012172013521,
      "num_tokens": 93924317.0,
      "step": 3510
    },
    {
      "entropy": 0.09190650654491037,
      "epoch": 6.263491710248136,
      "grad_norm": 1.234375,
      "learning_rate": 1.4402222222222224e-05,
      "loss": 0.0556,
      "mean_token_accuracy": 0.9847326755523682,
      "num_tokens": 94196164.0,
      "step": 3520
    },
    {
      "entropy": 0.09386124082375318,
      "epoch": 6.281295204183821,
      "grad_norm": 1.3515625,
      "learning_rate": 1.4380000000000001e-05,
      "loss": 0.0566,
      "mean_token_accuracy": 0.9846146836876869,
      "num_tokens": 94457901.0,
      "step": 3530
    },
    {
      "entropy": 0.09157110038213431,
      "epoch": 6.299098698119506,
      "grad_norm": 1.2109375,
      "learning_rate": 1.4357777777777779e-05,
      "loss": 0.0551,
      "mean_token_accuracy": 0.985302485153079,
      "num_tokens": 94731740.0,
      "step": 3540
    },
    {
      "entropy": 0.09264551640953869,
      "epoch": 6.316902192055191,
      "grad_norm": 1.2890625,
      "learning_rate": 1.4335555555555556e-05,
      "loss": 0.0567,
      "mean_token_accuracy": 0.9846193309873342,
      "num_tokens": 95001091.0,
      "step": 3550
    },
    {
      "entropy": 0.09399516386911273,
      "epoch": 6.334705685990876,
      "grad_norm": 1.4140625,
      "learning_rate": 1.4313333333333334e-05,
      "loss": 0.0567,
      "mean_token_accuracy": 0.9848186518996954,
      "num_tokens": 95262521.0,
      "step": 3560
    },
    {
      "entropy": 0.09336751853115857,
      "epoch": 6.35250917992656,
      "grad_norm": 1.34375,
      "learning_rate": 1.4291111111111113e-05,
      "loss": 0.057,
      "mean_token_accuracy": 0.9845899634063244,
      "num_tokens": 95532646.0,
      "step": 3570
    },
    {
      "entropy": 0.09054642412811517,
      "epoch": 6.370312673862245,
      "grad_norm": 1.140625,
      "learning_rate": 1.426888888888889e-05,
      "loss": 0.0554,
      "mean_token_accuracy": 0.9849374942481518,
      "num_tokens": 95800475.0,
      "step": 3580
    },
    {
      "entropy": 0.0928717106115073,
      "epoch": 6.38811616779793,
      "grad_norm": 1.3828125,
      "learning_rate": 1.4246666666666669e-05,
      "loss": 0.0567,
      "mean_token_accuracy": 0.9845525044947863,
      "num_tokens": 96072838.0,
      "step": 3590
    },
    {
      "entropy": 0.09396014239173382,
      "epoch": 6.405919661733615,
      "grad_norm": 1.078125,
      "learning_rate": 1.4224444444444445e-05,
      "loss": 0.0576,
      "mean_token_accuracy": 0.9842600673437119,
      "num_tokens": 96337796.0,
      "step": 3600
    },
    {
      "epoch": 6.405919661733615,
      "eval_chemistry_entropy": 0.2318279058933258,
      "eval_chemistry_loss": 1.1238375902175903,
      "eval_chemistry_mean_token_accuracy": 0.8093496389389038,
      "eval_chemistry_num_tokens": 96337796.0,
      "eval_chemistry_runtime": 34.6339,
      "eval_chemistry_samples_per_second": 14.437,
      "eval_chemistry_steps_per_second": 3.609,
      "step": 3600
    },
    {
      "epoch": 6.405919661733615,
      "eval_physics_entropy": 0.22755817019939423,
      "eval_physics_loss": 1.162530779838562,
      "eval_physics_mean_token_accuracy": 0.8043251366615295,
      "eval_physics_num_tokens": 96337796.0,
      "eval_physics_runtime": 41.2519,
      "eval_physics_samples_per_second": 12.121,
      "eval_physics_steps_per_second": 3.03,
      "step": 3600
    },
    {
      "entropy": 0.093810392357409,
      "epoch": 6.4237231556693,
      "grad_norm": 1.3828125,
      "learning_rate": 1.4202222222222224e-05,
      "loss": 0.0569,
      "mean_token_accuracy": 0.9847653720527887,
      "num_tokens": 96605272.0,
      "step": 3610
    },
    {
      "entropy": 0.09540869938209653,
      "epoch": 6.4415266496049846,
      "grad_norm": 1.171875,
      "learning_rate": 1.418e-05,
      "loss": 0.0586,
      "mean_token_accuracy": 0.9844035148620606,
      "num_tokens": 96861016.0,
      "step": 3620
    },
    {
      "entropy": 0.09337249633390457,
      "epoch": 6.45933014354067,
      "grad_norm": 1.6015625,
      "learning_rate": 1.4157777777777779e-05,
      "loss": 0.0571,
      "mean_token_accuracy": 0.984458052739501,
      "num_tokens": 97131274.0,
      "step": 3630
    },
    {
      "entropy": 0.09218331982847303,
      "epoch": 6.477133637476355,
      "grad_norm": 1.1484375,
      "learning_rate": 1.4135555555555555e-05,
      "loss": 0.0555,
      "mean_token_accuracy": 0.9848512411117554,
      "num_tokens": 97407303.0,
      "step": 3640
    },
    {
      "entropy": 0.09213897732552141,
      "epoch": 6.49493713141204,
      "grad_norm": 1.375,
      "learning_rate": 1.4113333333333334e-05,
      "loss": 0.0567,
      "mean_token_accuracy": 0.9847160268574953,
      "num_tokens": 97679446.0,
      "step": 3650
    },
    {
      "entropy": 0.0930451259482652,
      "epoch": 6.512740625347725,
      "grad_norm": 1.4453125,
      "learning_rate": 1.4091111111111114e-05,
      "loss": 0.0567,
      "mean_token_accuracy": 0.9847344007343054,
      "num_tokens": 97949047.0,
      "step": 3660
    },
    {
      "entropy": 0.0935262572253123,
      "epoch": 6.53054411928341,
      "grad_norm": 1.2578125,
      "learning_rate": 1.406888888888889e-05,
      "loss": 0.0566,
      "mean_token_accuracy": 0.9846566285938024,
      "num_tokens": 98216673.0,
      "step": 3670
    },
    {
      "entropy": 0.09259770622011274,
      "epoch": 6.548347613219094,
      "grad_norm": 1.203125,
      "learning_rate": 1.4046666666666669e-05,
      "loss": 0.0566,
      "mean_token_accuracy": 0.9847495544701814,
      "num_tokens": 98488147.0,
      "step": 3680
    },
    {
      "entropy": 0.09370218033436686,
      "epoch": 6.566151107154779,
      "grad_norm": 1.234375,
      "learning_rate": 1.4024444444444445e-05,
      "loss": 0.0577,
      "mean_token_accuracy": 0.9844383250921964,
      "num_tokens": 98751858.0,
      "step": 3690
    },
    {
      "entropy": 0.09372138518374414,
      "epoch": 6.583954601090464,
      "grad_norm": 1.328125,
      "learning_rate": 1.4002222222222224e-05,
      "loss": 0.0567,
      "mean_token_accuracy": 0.9845237113535404,
      "num_tokens": 99016531.0,
      "step": 3700
    },
    {
      "epoch": 6.583954601090464,
      "eval_chemistry_entropy": 0.22970075368881226,
      "eval_chemistry_loss": 1.1306341886520386,
      "eval_chemistry_mean_token_accuracy": 0.8095383520126342,
      "eval_chemistry_num_tokens": 99016531.0,
      "eval_chemistry_runtime": 34.9759,
      "eval_chemistry_samples_per_second": 14.296,
      "eval_chemistry_steps_per_second": 3.574,
      "step": 3700
    },
    {
      "epoch": 6.583954601090464,
      "eval_physics_entropy": 0.22417141306400298,
      "eval_physics_loss": 1.1724853515625,
      "eval_physics_mean_token_accuracy": 0.8047308712005615,
      "eval_physics_num_tokens": 99016531.0,
      "eval_physics_runtime": 41.2677,
      "eval_physics_samples_per_second": 12.116,
      "eval_physics_steps_per_second": 3.029,
      "step": 3700
    },
    {
      "entropy": 0.0929783372906968,
      "epoch": 6.601758095026149,
      "grad_norm": 1.2265625,
      "learning_rate": 1.398e-05,
      "loss": 0.0559,
      "mean_token_accuracy": 0.9850175462663173,
      "num_tokens": 99285898.0,
      "step": 3710
    },
    {
      "entropy": 0.09287062482908368,
      "epoch": 6.619561588961834,
      "grad_norm": 1.359375,
      "learning_rate": 1.395777777777778e-05,
      "loss": 0.0559,
      "mean_token_accuracy": 0.9851460207253695,
      "num_tokens": 99559113.0,
      "step": 3720
    },
    {
      "entropy": 0.09595338846556842,
      "epoch": 6.637365082897519,
      "grad_norm": 1.3203125,
      "learning_rate": 1.3935555555555557e-05,
      "loss": 0.0579,
      "mean_token_accuracy": 0.9843364715576172,
      "num_tokens": 99821848.0,
      "step": 3730
    },
    {
      "entropy": 0.09241803083568811,
      "epoch": 6.655168576833203,
      "grad_norm": 1.28125,
      "learning_rate": 1.3913333333333335e-05,
      "loss": 0.0578,
      "mean_token_accuracy": 0.9844335827976465,
      "num_tokens": 100091053.0,
      "step": 3740
    },
    {
      "entropy": 0.09336814917623996,
      "epoch": 6.672972070768888,
      "grad_norm": 1.5078125,
      "learning_rate": 1.3891111111111114e-05,
      "loss": 0.0572,
      "mean_token_accuracy": 0.9844303354620934,
      "num_tokens": 100363066.0,
      "step": 3750
    },
    {
      "entropy": 0.09491718888748438,
      "epoch": 6.690775564704573,
      "grad_norm": 1.3515625,
      "learning_rate": 1.386888888888889e-05,
      "loss": 0.0579,
      "mean_token_accuracy": 0.9847197256982326,
      "num_tokens": 100627481.0,
      "step": 3760
    },
    {
      "entropy": 0.09155842601321637,
      "epoch": 6.708579058640258,
      "grad_norm": 1.2734375,
      "learning_rate": 1.3846666666666669e-05,
      "loss": 0.0556,
      "mean_token_accuracy": 0.9846737906336784,
      "num_tokens": 100897551.0,
      "step": 3770
    },
    {
      "entropy": 0.09351664285641163,
      "epoch": 6.726382552575943,
      "grad_norm": 1.1875,
      "learning_rate": 1.3824444444444445e-05,
      "loss": 0.0569,
      "mean_token_accuracy": 0.9846763335168361,
      "num_tokens": 101169026.0,
      "step": 3780
    },
    {
      "entropy": 0.09503211197443306,
      "epoch": 6.7441860465116275,
      "grad_norm": 1.203125,
      "learning_rate": 1.3802222222222224e-05,
      "loss": 0.0591,
      "mean_token_accuracy": 0.984011958539486,
      "num_tokens": 101434647.0,
      "step": 3790
    },
    {
      "entropy": 0.09358547409065068,
      "epoch": 6.7619895404473125,
      "grad_norm": 1.125,
      "learning_rate": 1.378e-05,
      "loss": 0.0559,
      "mean_token_accuracy": 0.9849576093256474,
      "num_tokens": 101706355.0,
      "step": 3800
    },
    {
      "epoch": 6.7619895404473125,
      "eval_chemistry_entropy": 0.22803608918190002,
      "eval_chemistry_loss": 1.1343156099319458,
      "eval_chemistry_mean_token_accuracy": 0.8096190657615662,
      "eval_chemistry_num_tokens": 101706355.0,
      "eval_chemistry_runtime": 34.6121,
      "eval_chemistry_samples_per_second": 14.446,
      "eval_chemistry_steps_per_second": 3.611,
      "step": 3800
    },
    {
      "epoch": 6.7619895404473125,
      "eval_physics_entropy": 0.2231332515478134,
      "eval_physics_loss": 1.1767140626907349,
      "eval_physics_mean_token_accuracy": 0.8043256916999817,
      "eval_physics_num_tokens": 101706355.0,
      "eval_physics_runtime": 41.2688,
      "eval_physics_samples_per_second": 12.116,
      "eval_physics_steps_per_second": 3.029,
      "step": 3800
    },
    {
      "entropy": 0.09278706095647067,
      "epoch": 6.7797930343829975,
      "grad_norm": 1.3359375,
      "learning_rate": 1.375777777777778e-05,
      "loss": 0.0579,
      "mean_token_accuracy": 0.9843939192593097,
      "num_tokens": 101968752.0,
      "step": 3810
    },
    {
      "entropy": 0.09254231827799231,
      "epoch": 6.7975965283186826,
      "grad_norm": 1.28125,
      "learning_rate": 1.3735555555555557e-05,
      "loss": 0.0567,
      "mean_token_accuracy": 0.984549593925476,
      "num_tokens": 102238534.0,
      "step": 3820
    },
    {
      "entropy": 0.09356112710665912,
      "epoch": 6.815400022254368,
      "grad_norm": 1.2421875,
      "learning_rate": 1.3713333333333335e-05,
      "loss": 0.057,
      "mean_token_accuracy": 0.9846817117184401,
      "num_tokens": 102498964.0,
      "step": 3830
    },
    {
      "entropy": 0.09393304684199393,
      "epoch": 6.833203516190053,
      "grad_norm": 1.25,
      "learning_rate": 1.3691111111111112e-05,
      "loss": 0.0583,
      "mean_token_accuracy": 0.9840495180338621,
      "num_tokens": 102762816.0,
      "step": 3840
    },
    {
      "entropy": 0.09366672572214156,
      "epoch": 6.851007010125738,
      "grad_norm": 1.3203125,
      "learning_rate": 1.366888888888889e-05,
      "loss": 0.0574,
      "mean_token_accuracy": 0.9842468407005072,
      "num_tokens": 103029537.0,
      "step": 3850
    },
    {
      "entropy": 0.09258485166355968,
      "epoch": 6.868810504061422,
      "grad_norm": 1.1484375,
      "learning_rate": 1.3646666666666668e-05,
      "loss": 0.0576,
      "mean_token_accuracy": 0.9846377894282341,
      "num_tokens": 103301281.0,
      "step": 3860
    },
    {
      "entropy": 0.09303798568435014,
      "epoch": 6.886613997997107,
      "grad_norm": 1.28125,
      "learning_rate": 1.3624444444444445e-05,
      "loss": 0.0575,
      "mean_token_accuracy": 0.9843727514147759,
      "num_tokens": 103567835.0,
      "step": 3870
    },
    {
      "entropy": 0.09548078859224915,
      "epoch": 6.904417491932792,
      "grad_norm": 1.2734375,
      "learning_rate": 1.3602222222222223e-05,
      "loss": 0.0582,
      "mean_token_accuracy": 0.9841903056949377,
      "num_tokens": 103829891.0,
      "step": 3880
    },
    {
      "entropy": 0.09104312681593,
      "epoch": 6.922220985868477,
      "grad_norm": 1.40625,
      "learning_rate": 1.3580000000000002e-05,
      "loss": 0.0564,
      "mean_token_accuracy": 0.9848036479204894,
      "num_tokens": 104101701.0,
      "step": 3890
    },
    {
      "entropy": 0.09530101537238807,
      "epoch": 6.940024479804162,
      "grad_norm": 1.3203125,
      "learning_rate": 1.3557777777777778e-05,
      "loss": 0.0581,
      "mean_token_accuracy": 0.9839477833360434,
      "num_tokens": 104362319.0,
      "step": 3900
    },
    {
      "epoch": 6.940024479804162,
      "eval_chemistry_entropy": 0.22867212694883346,
      "eval_chemistry_loss": 1.1353645324707031,
      "eval_chemistry_mean_token_accuracy": 0.8095029134750367,
      "eval_chemistry_num_tokens": 104362319.0,
      "eval_chemistry_runtime": 34.6346,
      "eval_chemistry_samples_per_second": 14.436,
      "eval_chemistry_steps_per_second": 3.609,
      "step": 3900
    },
    {
      "epoch": 6.940024479804162,
      "eval_physics_entropy": 0.22366387474536895,
      "eval_physics_loss": 1.1753313541412354,
      "eval_physics_mean_token_accuracy": 0.8048194098472595,
      "eval_physics_num_tokens": 104362319.0,
      "eval_physics_runtime": 41.26,
      "eval_physics_samples_per_second": 12.118,
      "eval_physics_steps_per_second": 3.03,
      "step": 3900
    },
    {
      "entropy": 0.0923372016986832,
      "epoch": 6.957827973739846,
      "grad_norm": 1.1875,
      "learning_rate": 1.3535555555555557e-05,
      "loss": 0.0558,
      "mean_token_accuracy": 0.9849498894065618,
      "num_tokens": 104633877.0,
      "step": 3910
    },
    {
      "entropy": 0.09448046216275544,
      "epoch": 6.975631467675531,
      "grad_norm": 1.359375,
      "learning_rate": 1.3513333333333333e-05,
      "loss": 0.0583,
      "mean_token_accuracy": 0.9842988450080157,
      "num_tokens": 104904851.0,
      "step": 3920
    },
    {
      "entropy": 0.09370976577047259,
      "epoch": 6.993434961611216,
      "grad_norm": 1.25,
      "learning_rate": 1.3491111111111113e-05,
      "loss": 0.0563,
      "mean_token_accuracy": 0.9845016706734896,
      "num_tokens": 105175455.0,
      "step": 3930
    },
    {
      "entropy": 0.08417318484956218,
      "epoch": 7.0106820963614105,
      "grad_norm": 0.87109375,
      "learning_rate": 1.3468888888888888e-05,
      "loss": 0.0488,
      "mean_token_accuracy": 0.9867334100507921,
      "num_tokens": 105430616.0,
      "step": 3940
    },
    {
      "entropy": 0.07191569162532688,
      "epoch": 7.0284855902970955,
      "grad_norm": 0.984375,
      "learning_rate": 1.3446666666666668e-05,
      "loss": 0.0424,
      "mean_token_accuracy": 0.9883563961833716,
      "num_tokens": 105700227.0,
      "step": 3950
    },
    {
      "entropy": 0.07360758790746331,
      "epoch": 7.0462890842327806,
      "grad_norm": 0.98828125,
      "learning_rate": 1.3424444444444447e-05,
      "loss": 0.0436,
      "mean_token_accuracy": 0.9878683175891638,
      "num_tokens": 105961486.0,
      "step": 3960
    },
    {
      "entropy": 0.07348821454215795,
      "epoch": 7.064092578168466,
      "grad_norm": 0.8828125,
      "learning_rate": 1.3402222222222223e-05,
      "loss": 0.0434,
      "mean_token_accuracy": 0.9881870541721582,
      "num_tokens": 106224145.0,
      "step": 3970
    },
    {
      "entropy": 0.07323711377102882,
      "epoch": 7.081896072104151,
      "grad_norm": 0.94921875,
      "learning_rate": 1.3380000000000002e-05,
      "loss": 0.0433,
      "mean_token_accuracy": 0.9880886465311051,
      "num_tokens": 106490616.0,
      "step": 3980
    },
    {
      "entropy": 0.07291189294774085,
      "epoch": 7.099699566039836,
      "grad_norm": 0.890625,
      "learning_rate": 1.3357777777777778e-05,
      "loss": 0.0439,
      "mean_token_accuracy": 0.9880160391330719,
      "num_tokens": 106749838.0,
      "step": 3990
    },
    {
      "entropy": 0.0719311147229746,
      "epoch": 7.11750305997552,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3335555555555558e-05,
      "loss": 0.0427,
      "mean_token_accuracy": 0.9881735738366842,
      "num_tokens": 107023362.0,
      "step": 4000
    },
    {
      "epoch": 7.11750305997552,
      "eval_chemistry_entropy": 0.2165147470831871,
      "eval_chemistry_loss": 1.1778162717819214,
      "eval_chemistry_mean_token_accuracy": 0.8104037828445435,
      "eval_chemistry_num_tokens": 107023362.0,
      "eval_chemistry_runtime": 34.6298,
      "eval_chemistry_samples_per_second": 14.438,
      "eval_chemistry_steps_per_second": 3.61,
      "step": 4000
    },
    {
      "epoch": 7.11750305997552,
      "eval_physics_entropy": 0.2114391126036644,
      "eval_physics_loss": 1.2207711935043335,
      "eval_physics_mean_token_accuracy": 0.8051733541488647,
      "eval_physics_num_tokens": 107023362.0,
      "eval_physics_runtime": 40.9434,
      "eval_physics_samples_per_second": 12.212,
      "eval_physics_steps_per_second": 3.053,
      "step": 4000
    },
    {
      "entropy": 0.07310699296649545,
      "epoch": 7.135306553911205,
      "grad_norm": 0.9140625,
      "learning_rate": 1.3313333333333333e-05,
      "loss": 0.0434,
      "mean_token_accuracy": 0.9881420850753784,
      "num_tokens": 107292798.0,
      "step": 4010
    },
    {
      "entropy": 0.07121766875497997,
      "epoch": 7.15311004784689,
      "grad_norm": 0.84765625,
      "learning_rate": 1.3291111111111113e-05,
      "loss": 0.0436,
      "mean_token_accuracy": 0.9883767809718847,
      "num_tokens": 107564764.0,
      "step": 4020
    },
    {
      "entropy": 0.0739193097455427,
      "epoch": 7.170913541782575,
      "grad_norm": 0.9375,
      "learning_rate": 1.3268888888888889e-05,
      "loss": 0.0431,
      "mean_token_accuracy": 0.9881718154996634,
      "num_tokens": 107838170.0,
      "step": 4030
    },
    {
      "entropy": 0.07300531500950455,
      "epoch": 7.18871703571826,
      "grad_norm": 1.046875,
      "learning_rate": 1.3246666666666668e-05,
      "loss": 0.0431,
      "mean_token_accuracy": 0.9881930593401194,
      "num_tokens": 108109623.0,
      "step": 4040
    },
    {
      "entropy": 0.07570609436370432,
      "epoch": 7.206520529653945,
      "grad_norm": 0.875,
      "learning_rate": 1.3224444444444446e-05,
      "loss": 0.0458,
      "mean_token_accuracy": 0.9873250342905522,
      "num_tokens": 108365234.0,
      "step": 4050
    },
    {
      "entropy": 0.07341724331490695,
      "epoch": 7.224324023589629,
      "grad_norm": 0.8125,
      "learning_rate": 1.3202222222222223e-05,
      "loss": 0.0442,
      "mean_token_accuracy": 0.9878484208136797,
      "num_tokens": 108632098.0,
      "step": 4060
    },
    {
      "entropy": 0.0727812558412552,
      "epoch": 7.242127517525314,
      "grad_norm": 0.734375,
      "learning_rate": 1.3180000000000001e-05,
      "loss": 0.0439,
      "mean_token_accuracy": 0.9878760565072298,
      "num_tokens": 108906557.0,
      "step": 4070
    },
    {
      "entropy": 0.07455522243399174,
      "epoch": 7.259931011460999,
      "grad_norm": 0.93359375,
      "learning_rate": 1.3157777777777778e-05,
      "loss": 0.0451,
      "mean_token_accuracy": 0.9876507122069598,
      "num_tokens": 109170243.0,
      "step": 4080
    },
    {
      "entropy": 0.07301900046877563,
      "epoch": 7.277734505396684,
      "grad_norm": 0.97265625,
      "learning_rate": 1.3135555555555558e-05,
      "loss": 0.0442,
      "mean_token_accuracy": 0.987964203581214,
      "num_tokens": 109433380.0,
      "step": 4090
    },
    {
      "entropy": 0.07449646205641329,
      "epoch": 7.295537999332369,
      "grad_norm": 0.91796875,
      "learning_rate": 1.3113333333333334e-05,
      "loss": 0.0447,
      "mean_token_accuracy": 0.9872370876371861,
      "num_tokens": 109697045.0,
      "step": 4100
    },
    {
      "epoch": 7.295537999332369,
      "eval_chemistry_entropy": 0.21526601415872573,
      "eval_chemistry_loss": 1.183851718902588,
      "eval_chemistry_mean_token_accuracy": 0.8104252300262451,
      "eval_chemistry_num_tokens": 109697045.0,
      "eval_chemistry_runtime": 34.5803,
      "eval_chemistry_samples_per_second": 14.459,
      "eval_chemistry_steps_per_second": 3.615,
      "step": 4100
    },
    {
      "epoch": 7.295537999332369,
      "eval_physics_entropy": 0.21061576992273331,
      "eval_physics_loss": 1.2267844676971436,
      "eval_physics_mean_token_accuracy": 0.8048838820457459,
      "eval_physics_num_tokens": 109697045.0,
      "eval_physics_runtime": 41.2485,
      "eval_physics_samples_per_second": 12.122,
      "eval_physics_steps_per_second": 3.03,
      "step": 4100
    },
    {
      "entropy": 0.0727313372073695,
      "epoch": 7.313341493268053,
      "grad_norm": 0.96875,
      "learning_rate": 1.3091111111111113e-05,
      "loss": 0.0435,
      "mean_token_accuracy": 0.9878450352698565,
      "num_tokens": 109964874.0,
      "step": 4110
    },
    {
      "entropy": 0.0728560226270929,
      "epoch": 7.331144987203738,
      "grad_norm": 0.9609375,
      "learning_rate": 1.306888888888889e-05,
      "loss": 0.0436,
      "mean_token_accuracy": 0.9879323188215494,
      "num_tokens": 110233835.0,
      "step": 4120
    },
    {
      "entropy": 0.07108670368324965,
      "epoch": 7.3489484811394234,
      "grad_norm": 0.83203125,
      "learning_rate": 1.3046666666666668e-05,
      "loss": 0.0423,
      "mean_token_accuracy": 0.9882055234164,
      "num_tokens": 110509871.0,
      "step": 4130
    },
    {
      "entropy": 0.07574002894107253,
      "epoch": 7.3667519750751085,
      "grad_norm": 0.95703125,
      "learning_rate": 1.3024444444444446e-05,
      "loss": 0.0452,
      "mean_token_accuracy": 0.9876693092286587,
      "num_tokens": 110774889.0,
      "step": 4140
    },
    {
      "entropy": 0.07414197595790029,
      "epoch": 7.3845554690107935,
      "grad_norm": 0.94140625,
      "learning_rate": 1.3002222222222223e-05,
      "loss": 0.0441,
      "mean_token_accuracy": 0.9876979842782021,
      "num_tokens": 111046685.0,
      "step": 4150
    },
    {
      "entropy": 0.07478623117785901,
      "epoch": 7.4023589629464785,
      "grad_norm": 0.9921875,
      "learning_rate": 1.2980000000000001e-05,
      "loss": 0.0453,
      "mean_token_accuracy": 0.9873840816318988,
      "num_tokens": 111309080.0,
      "step": 4160
    },
    {
      "entropy": 0.07402537071611733,
      "epoch": 7.420162456882164,
      "grad_norm": 0.90234375,
      "learning_rate": 1.2957777777777779e-05,
      "loss": 0.0452,
      "mean_token_accuracy": 0.9872989296913147,
      "num_tokens": 111571958.0,
      "step": 4170
    },
    {
      "entropy": 0.07201829818077385,
      "epoch": 7.437965950817848,
      "grad_norm": 1.0,
      "learning_rate": 1.2935555555555556e-05,
      "loss": 0.0441,
      "mean_token_accuracy": 0.9877836029976607,
      "num_tokens": 111845576.0,
      "step": 4180
    },
    {
      "entropy": 0.07186442227102816,
      "epoch": 7.455769444753533,
      "grad_norm": 0.88671875,
      "learning_rate": 1.2913333333333336e-05,
      "loss": 0.0442,
      "mean_token_accuracy": 0.9878892824053764,
      "num_tokens": 112115213.0,
      "step": 4190
    },
    {
      "entropy": 0.07595633622258902,
      "epoch": 7.473572938689218,
      "grad_norm": 1.140625,
      "learning_rate": 1.2891111111111112e-05,
      "loss": 0.0461,
      "mean_token_accuracy": 0.987379552423954,
      "num_tokens": 112375759.0,
      "step": 4200
    },
    {
      "epoch": 7.473572938689218,
      "eval_chemistry_entropy": 0.21608789557218552,
      "eval_chemistry_loss": 1.18284010887146,
      "eval_chemistry_mean_token_accuracy": 0.8102615876197815,
      "eval_chemistry_num_tokens": 112375759.0,
      "eval_chemistry_runtime": 34.6099,
      "eval_chemistry_samples_per_second": 14.447,
      "eval_chemistry_steps_per_second": 3.612,
      "step": 4200
    },
    {
      "epoch": 7.473572938689218,
      "eval_physics_entropy": 0.2111558927297592,
      "eval_physics_loss": 1.2275816202163696,
      "eval_physics_mean_token_accuracy": 0.8047467660903931,
      "eval_physics_num_tokens": 112375759.0,
      "eval_physics_runtime": 41.2332,
      "eval_physics_samples_per_second": 12.126,
      "eval_physics_steps_per_second": 3.032,
      "step": 4200
    },
    {
      "entropy": 0.0716058092424646,
      "epoch": 7.491376432624903,
      "grad_norm": 0.87109375,
      "learning_rate": 1.2868888888888891e-05,
      "loss": 0.043,
      "mean_token_accuracy": 0.9883722405880689,
      "num_tokens": 112648536.0,
      "step": 4210
    },
    {
      "entropy": 0.0755084725562483,
      "epoch": 7.509179926560588,
      "grad_norm": 0.96484375,
      "learning_rate": 1.2846666666666667e-05,
      "loss": 0.0452,
      "mean_token_accuracy": 0.9874132890254259,
      "num_tokens": 112909264.0,
      "step": 4220
    },
    {
      "entropy": 0.07487286000978202,
      "epoch": 7.526983420496272,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2824444444444446e-05,
      "loss": 0.0453,
      "mean_token_accuracy": 0.9876388993114233,
      "num_tokens": 113173577.0,
      "step": 4230
    },
    {
      "entropy": 0.07264250102452934,
      "epoch": 7.544786914431957,
      "grad_norm": 0.87109375,
      "learning_rate": 1.2802222222222222e-05,
      "loss": 0.044,
      "mean_token_accuracy": 0.9874496996402741,
      "num_tokens": 113444280.0,
      "step": 4240
    },
    {
      "entropy": 0.07303144540637732,
      "epoch": 7.562590408367642,
      "grad_norm": 0.9765625,
      "learning_rate": 1.2780000000000001e-05,
      "loss": 0.0444,
      "mean_token_accuracy": 0.9877622861415147,
      "num_tokens": 113710191.0,
      "step": 4250
    },
    {
      "entropy": 0.07486866011749953,
      "epoch": 7.580393902303327,
      "grad_norm": 0.84765625,
      "learning_rate": 1.2757777777777777e-05,
      "loss": 0.0443,
      "mean_token_accuracy": 0.9876678586006165,
      "num_tokens": 113976979.0,
      "step": 4260
    },
    {
      "entropy": 0.07296732109971345,
      "epoch": 7.598197396239012,
      "grad_norm": 0.98828125,
      "learning_rate": 1.2735555555555557e-05,
      "loss": 0.0436,
      "mean_token_accuracy": 0.9879147756844759,
      "num_tokens": 114248096.0,
      "step": 4270
    },
    {
      "entropy": 0.07390863625332714,
      "epoch": 7.616000890174697,
      "grad_norm": 0.84375,
      "learning_rate": 1.2713333333333336e-05,
      "loss": 0.0442,
      "mean_token_accuracy": 0.9879091136157513,
      "num_tokens": 114514925.0,
      "step": 4280
    },
    {
      "entropy": 0.07360817035660147,
      "epoch": 7.633804384110381,
      "grad_norm": 1.3984375,
      "learning_rate": 1.2691111111111112e-05,
      "loss": 0.0455,
      "mean_token_accuracy": 0.9871931057423353,
      "num_tokens": 114778978.0,
      "step": 4290
    },
    {
      "entropy": 0.07449810754042119,
      "epoch": 7.651607878046066,
      "grad_norm": 0.96875,
      "learning_rate": 1.2668888888888891e-05,
      "loss": 0.045,
      "mean_token_accuracy": 0.9874308243393898,
      "num_tokens": 115044692.0,
      "step": 4300
    },
    {
      "epoch": 7.651607878046066,
      "eval_chemistry_entropy": 0.2154428989291191,
      "eval_chemistry_loss": 1.188363790512085,
      "eval_chemistry_mean_token_accuracy": 0.8103943915367127,
      "eval_chemistry_num_tokens": 115044692.0,
      "eval_chemistry_runtime": 34.5862,
      "eval_chemistry_samples_per_second": 14.457,
      "eval_chemistry_steps_per_second": 3.614,
      "step": 4300
    },
    {
      "epoch": 7.651607878046066,
      "eval_physics_entropy": 0.21011914831399917,
      "eval_physics_loss": 1.2344199419021606,
      "eval_physics_mean_token_accuracy": 0.8048519930839538,
      "eval_physics_num_tokens": 115044692.0,
      "eval_physics_runtime": 41.2726,
      "eval_physics_samples_per_second": 12.115,
      "eval_physics_steps_per_second": 3.029,
      "step": 4300
    },
    {
      "entropy": 0.07511928507592529,
      "epoch": 7.669411371981751,
      "grad_norm": 1.21875,
      "learning_rate": 1.2646666666666667e-05,
      "loss": 0.0455,
      "mean_token_accuracy": 0.987170723080635,
      "num_tokens": 115305912.0,
      "step": 4310
    },
    {
      "entropy": 0.07125148598570377,
      "epoch": 7.687214865917436,
      "grad_norm": 0.93359375,
      "learning_rate": 1.2624444444444446e-05,
      "loss": 0.0431,
      "mean_token_accuracy": 0.9881943237036467,
      "num_tokens": 115585713.0,
      "step": 4320
    },
    {
      "entropy": 0.0741977678379044,
      "epoch": 7.7050183598531214,
      "grad_norm": 0.88671875,
      "learning_rate": 1.2602222222222222e-05,
      "loss": 0.0444,
      "mean_token_accuracy": 0.9877401128411293,
      "num_tokens": 115857231.0,
      "step": 4330
    },
    {
      "entropy": 0.07447677634190768,
      "epoch": 7.7228218537888065,
      "grad_norm": 0.890625,
      "learning_rate": 1.2580000000000002e-05,
      "loss": 0.0449,
      "mean_token_accuracy": 0.9873528491705656,
      "num_tokens": 116122213.0,
      "step": 4340
    },
    {
      "entropy": 0.0715603593736887,
      "epoch": 7.740625347724491,
      "grad_norm": 0.87890625,
      "learning_rate": 1.2557777777777779e-05,
      "loss": 0.044,
      "mean_token_accuracy": 0.9880015090107918,
      "num_tokens": 116395012.0,
      "step": 4350
    },
    {
      "entropy": 0.07471361048519612,
      "epoch": 7.758428841660176,
      "grad_norm": 0.88671875,
      "learning_rate": 1.2535555555555557e-05,
      "loss": 0.0445,
      "mean_token_accuracy": 0.9873521786183119,
      "num_tokens": 116662588.0,
      "step": 4360
    },
    {
      "entropy": 0.07219609864987434,
      "epoch": 7.776232335595861,
      "grad_norm": 0.94921875,
      "learning_rate": 1.2513333333333334e-05,
      "loss": 0.0437,
      "mean_token_accuracy": 0.9879470851272345,
      "num_tokens": 116938754.0,
      "step": 4370
    },
    {
      "entropy": 0.07468674255069345,
      "epoch": 7.794035829531546,
      "grad_norm": 1.09375,
      "learning_rate": 1.2491111111111112e-05,
      "loss": 0.0448,
      "mean_token_accuracy": 0.9875706970691681,
      "num_tokens": 117207686.0,
      "step": 4380
    },
    {
      "entropy": 0.07355934337247164,
      "epoch": 7.811839323467231,
      "grad_norm": 0.91796875,
      "learning_rate": 1.246888888888889e-05,
      "loss": 0.0449,
      "mean_token_accuracy": 0.987399535253644,
      "num_tokens": 117473530.0,
      "step": 4390
    },
    {
      "entropy": 0.07281780790071934,
      "epoch": 7.829642817402915,
      "grad_norm": 1.0390625,
      "learning_rate": 1.2446666666666667e-05,
      "loss": 0.0459,
      "mean_token_accuracy": 0.9875079352408648,
      "num_tokens": 117738945.0,
      "step": 4400
    },
    {
      "epoch": 7.829642817402915,
      "eval_chemistry_entropy": 0.21491128677129745,
      "eval_chemistry_loss": 1.1903232336044312,
      "eval_chemistry_mean_token_accuracy": 0.810541296005249,
      "eval_chemistry_num_tokens": 117738945.0,
      "eval_chemistry_runtime": 34.62,
      "eval_chemistry_samples_per_second": 14.443,
      "eval_chemistry_steps_per_second": 3.611,
      "step": 4400
    },
    {
      "epoch": 7.829642817402915,
      "eval_physics_entropy": 0.20959068912267684,
      "eval_physics_loss": 1.2364140748977661,
      "eval_physics_mean_token_accuracy": 0.8050095362663269,
      "eval_physics_num_tokens": 117738945.0,
      "eval_physics_runtime": 41.2322,
      "eval_physics_samples_per_second": 12.126,
      "eval_physics_steps_per_second": 3.032,
      "step": 4400
    },
    {
      "entropy": 0.07364364902023227,
      "epoch": 7.8474463113386,
      "grad_norm": 0.86328125,
      "learning_rate": 1.2424444444444445e-05,
      "loss": 0.0463,
      "mean_token_accuracy": 0.9872796233743429,
      "num_tokens": 118002841.0,
      "step": 4410
    },
    {
      "entropy": 0.07310810647904872,
      "epoch": 7.865249805274285,
      "grad_norm": 0.88671875,
      "learning_rate": 1.2402222222222222e-05,
      "loss": 0.0442,
      "mean_token_accuracy": 0.9878404460847378,
      "num_tokens": 118277625.0,
      "step": 4420
    },
    {
      "entropy": 0.07360601213295012,
      "epoch": 7.88305329920997,
      "grad_norm": 0.8984375,
      "learning_rate": 1.2380000000000002e-05,
      "loss": 0.0453,
      "mean_token_accuracy": 0.9872990295290947,
      "num_tokens": 118543382.0,
      "step": 4430
    },
    {
      "entropy": 0.07125959191471339,
      "epoch": 7.900856793145655,
      "grad_norm": 0.8671875,
      "learning_rate": 1.235777777777778e-05,
      "loss": 0.044,
      "mean_token_accuracy": 0.9879327766597271,
      "num_tokens": 118817598.0,
      "step": 4440
    },
    {
      "entropy": 0.0735878188861534,
      "epoch": 7.91866028708134,
      "grad_norm": 1.015625,
      "learning_rate": 1.2335555555555557e-05,
      "loss": 0.0438,
      "mean_token_accuracy": 0.9878993470221757,
      "num_tokens": 119088753.0,
      "step": 4450
    },
    {
      "entropy": 0.07451147942338139,
      "epoch": 7.936463781017024,
      "grad_norm": 0.9765625,
      "learning_rate": 1.2313333333333335e-05,
      "loss": 0.0448,
      "mean_token_accuracy": 0.9873662251979113,
      "num_tokens": 119355949.0,
      "step": 4460
    },
    {
      "entropy": 0.07238080403767526,
      "epoch": 7.954267274952709,
      "grad_norm": 0.9765625,
      "learning_rate": 1.2291111111111112e-05,
      "loss": 0.044,
      "mean_token_accuracy": 0.9875876381993294,
      "num_tokens": 119628627.0,
      "step": 4470
    },
    {
      "entropy": 0.07326499768532813,
      "epoch": 7.972070768888394,
      "grad_norm": 0.78515625,
      "learning_rate": 1.226888888888889e-05,
      "loss": 0.0445,
      "mean_token_accuracy": 0.9874961365014314,
      "num_tokens": 119898193.0,
      "step": 4480
    },
    {
      "entropy": 0.07430332913063467,
      "epoch": 7.989874262824079,
      "grad_norm": 0.8828125,
      "learning_rate": 1.2246666666666667e-05,
      "loss": 0.045,
      "mean_token_accuracy": 0.9874973658472299,
      "num_tokens": 120166302.0,
      "step": 4490
    },
    {
      "entropy": 0.07022783333255399,
      "epoch": 8.007121397574274,
      "grad_norm": 0.51171875,
      "learning_rate": 1.2224444444444445e-05,
      "loss": 0.0414,
      "mean_token_accuracy": 0.988403715625886,
      "num_tokens": 120427983.0,
      "step": 4500
    },
    {
      "epoch": 8.007121397574274,
      "eval_chemistry_entropy": 0.21411660689115525,
      "eval_chemistry_loss": 1.1947780847549438,
      "eval_chemistry_mean_token_accuracy": 0.8101702346801758,
      "eval_chemistry_num_tokens": 120427983.0,
      "eval_chemistry_runtime": 34.6129,
      "eval_chemistry_samples_per_second": 14.445,
      "eval_chemistry_steps_per_second": 3.611,
      "step": 4500
    },
    {
      "epoch": 8.007121397574274,
      "eval_physics_entropy": 0.20912677782773972,
      "eval_physics_loss": 1.2414851188659668,
      "eval_physics_mean_token_accuracy": 0.8045782704353333,
      "eval_physics_num_tokens": 120427983.0,
      "eval_physics_runtime": 40.9697,
      "eval_physics_samples_per_second": 12.204,
      "eval_physics_steps_per_second": 3.051,
      "step": 4500
    },
    {
      "entropy": 0.06578118347097188,
      "epoch": 8.02492489150996,
      "grad_norm": 0.54296875,
      "learning_rate": 1.2202222222222224e-05,
      "loss": 0.0376,
      "mean_token_accuracy": 0.9892934214323759,
      "num_tokens": 120695599.0,
      "step": 4510
    },
    {
      "entropy": 0.0610331998555921,
      "epoch": 8.042728385445644,
      "grad_norm": 0.5703125,
      "learning_rate": 1.218e-05,
      "loss": 0.0365,
      "mean_token_accuracy": 0.989518266543746,
      "num_tokens": 120969526.0,
      "step": 4520
    },
    {
      "entropy": 0.06369996154680849,
      "epoch": 8.060531879381328,
      "grad_norm": 0.67578125,
      "learning_rate": 1.215777777777778e-05,
      "loss": 0.0372,
      "mean_token_accuracy": 0.9893280819058419,
      "num_tokens": 121239327.0,
      "step": 4530
    },
    {
      "entropy": 0.06250840050633996,
      "epoch": 8.078335373317014,
      "grad_norm": 0.61328125,
      "learning_rate": 1.2135555555555556e-05,
      "loss": 0.0366,
      "mean_token_accuracy": 0.9896233890205621,
      "num_tokens": 121509021.0,
      "step": 4540
    },
    {
      "entropy": 0.06327347565675154,
      "epoch": 8.096138867252698,
      "grad_norm": 0.5859375,
      "learning_rate": 1.2113333333333335e-05,
      "loss": 0.0368,
      "mean_token_accuracy": 0.9891492255032063,
      "num_tokens": 121779298.0,
      "step": 4550
    },
    {
      "entropy": 0.06293636739719659,
      "epoch": 8.113942361188384,
      "grad_norm": 0.5078125,
      "learning_rate": 1.209111111111111e-05,
      "loss": 0.0368,
      "mean_token_accuracy": 0.9893124159425497,
      "num_tokens": 122047848.0,
      "step": 4560
    },
    {
      "entropy": 0.062493463954888286,
      "epoch": 8.131745855124068,
      "grad_norm": 0.49609375,
      "learning_rate": 1.206888888888889e-05,
      "loss": 0.0373,
      "mean_token_accuracy": 0.9893547147512436,
      "num_tokens": 122314669.0,
      "step": 4570
    },
    {
      "entropy": 0.06452988574746996,
      "epoch": 8.149549349059752,
      "grad_norm": 0.71484375,
      "learning_rate": 1.204666666666667e-05,
      "loss": 0.0374,
      "mean_token_accuracy": 0.9892832513898611,
      "num_tokens": 122580631.0,
      "step": 4580
    },
    {
      "entropy": 0.06271463485900312,
      "epoch": 8.167352842995438,
      "grad_norm": 0.494140625,
      "learning_rate": 1.2024444444444445e-05,
      "loss": 0.0374,
      "mean_token_accuracy": 0.9893312480300664,
      "num_tokens": 122846504.0,
      "step": 4590
    },
    {
      "entropy": 0.061985064297914505,
      "epoch": 8.185156336931122,
      "grad_norm": 0.61328125,
      "learning_rate": 1.2002222222222225e-05,
      "loss": 0.0369,
      "mean_token_accuracy": 0.9894137635827065,
      "num_tokens": 123117342.0,
      "step": 4600
    },
    {
      "epoch": 8.185156336931122,
      "eval_chemistry_entropy": 0.20670313078165053,
      "eval_chemistry_loss": 1.2225805521011353,
      "eval_chemistry_mean_token_accuracy": 0.8108696904182434,
      "eval_chemistry_num_tokens": 123117342.0,
      "eval_chemistry_runtime": 34.6631,
      "eval_chemistry_samples_per_second": 14.425,
      "eval_chemistry_steps_per_second": 3.606,
      "step": 4600
    },
    {
      "epoch": 8.185156336931122,
      "eval_physics_entropy": 0.202040593624115,
      "eval_physics_loss": 1.2721084356307983,
      "eval_physics_mean_token_accuracy": 0.8050774235725403,
      "eval_physics_num_tokens": 123117342.0,
      "eval_physics_runtime": 41.2957,
      "eval_physics_samples_per_second": 12.108,
      "eval_physics_steps_per_second": 3.027,
      "step": 4600
    },
    {
      "entropy": 0.06312562404200435,
      "epoch": 8.202959830866808,
      "grad_norm": 0.5390625,
      "learning_rate": 1.198e-05,
      "loss": 0.0373,
      "mean_token_accuracy": 0.9892840754240751,
      "num_tokens": 123386985.0,
      "step": 4610
    },
    {
      "entropy": 0.06247786502353847,
      "epoch": 8.220763324802492,
      "grad_norm": 0.53515625,
      "learning_rate": 1.195777777777778e-05,
      "loss": 0.0377,
      "mean_token_accuracy": 0.9892281461507082,
      "num_tokens": 123651828.0,
      "step": 4620
    },
    {
      "entropy": 0.05978459210600704,
      "epoch": 8.238566818738178,
      "grad_norm": 0.466796875,
      "learning_rate": 1.1935555555555556e-05,
      "loss": 0.0354,
      "mean_token_accuracy": 0.9899001281708479,
      "num_tokens": 123933897.0,
      "step": 4630
    },
    {
      "entropy": 0.06502155850175768,
      "epoch": 8.256370312673862,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1913333333333335e-05,
      "loss": 0.0382,
      "mean_token_accuracy": 0.9888288144022226,
      "num_tokens": 124199369.0,
      "step": 4640
    },
    {
      "entropy": 0.06480145251844079,
      "epoch": 8.274173806609546,
      "grad_norm": 0.52734375,
      "learning_rate": 1.1891111111111111e-05,
      "loss": 0.0388,
      "mean_token_accuracy": 0.9886323850601911,
      "num_tokens": 124455584.0,
      "step": 4650
    },
    {
      "entropy": 0.06222098821308464,
      "epoch": 8.291977300545232,
      "grad_norm": 0.57421875,
      "learning_rate": 1.186888888888889e-05,
      "loss": 0.0373,
      "mean_token_accuracy": 0.9892554465681315,
      "num_tokens": 124722324.0,
      "step": 4660
    },
    {
      "entropy": 0.06261638691648841,
      "epoch": 8.309780794480917,
      "grad_norm": 0.5078125,
      "learning_rate": 1.1846666666666668e-05,
      "loss": 0.0381,
      "mean_token_accuracy": 0.9891822930425406,
      "num_tokens": 124985365.0,
      "step": 4670
    },
    {
      "entropy": 0.06409567834343761,
      "epoch": 8.327584288416602,
      "grad_norm": 0.671875,
      "learning_rate": 1.1824444444444445e-05,
      "loss": 0.0381,
      "mean_token_accuracy": 0.9890421740710735,
      "num_tokens": 125248583.0,
      "step": 4680
    },
    {
      "entropy": 0.06370994239114225,
      "epoch": 8.345387782352287,
      "grad_norm": 0.60546875,
      "learning_rate": 1.1802222222222223e-05,
      "loss": 0.0376,
      "mean_token_accuracy": 0.9891662824898958,
      "num_tokens": 125514621.0,
      "step": 4690
    },
    {
      "entropy": 0.06408025913406164,
      "epoch": 8.36319127628797,
      "grad_norm": 0.53125,
      "learning_rate": 1.178e-05,
      "loss": 0.0389,
      "mean_token_accuracy": 0.9891206040978432,
      "num_tokens": 125774044.0,
      "step": 4700
    },
    {
      "epoch": 8.36319127628797,
      "eval_chemistry_entropy": 0.20522890150547027,
      "eval_chemistry_loss": 1.2274879217147827,
      "eval_chemistry_mean_token_accuracy": 0.8108414783477783,
      "eval_chemistry_num_tokens": 125774044.0,
      "eval_chemistry_runtime": 34.6461,
      "eval_chemistry_samples_per_second": 14.432,
      "eval_chemistry_steps_per_second": 3.608,
      "step": 4700
    },
    {
      "epoch": 8.36319127628797,
      "eval_physics_entropy": 0.20076260989904404,
      "eval_physics_loss": 1.2789329290390015,
      "eval_physics_mean_token_accuracy": 0.8052958278656006,
      "eval_physics_num_tokens": 125774044.0,
      "eval_physics_runtime": 41.2679,
      "eval_physics_samples_per_second": 12.116,
      "eval_physics_steps_per_second": 3.029,
      "step": 4700
    },
    {
      "entropy": 0.06281561562791467,
      "epoch": 8.380994770223657,
      "grad_norm": 0.62890625,
      "learning_rate": 1.1757777777777778e-05,
      "loss": 0.0381,
      "mean_token_accuracy": 0.9890597574412823,
      "num_tokens": 126039416.0,
      "step": 4710
    },
    {
      "entropy": 0.0628769415896386,
      "epoch": 8.39879826415934,
      "grad_norm": 0.734375,
      "learning_rate": 1.1735555555555556e-05,
      "loss": 0.0373,
      "mean_token_accuracy": 0.9892385397106409,
      "num_tokens": 126307871.0,
      "step": 4720
    },
    {
      "entropy": 0.06376641327515245,
      "epoch": 8.416601758095027,
      "grad_norm": 0.64453125,
      "learning_rate": 1.1713333333333334e-05,
      "loss": 0.0378,
      "mean_token_accuracy": 0.9890662416815758,
      "num_tokens": 126573591.0,
      "step": 4730
    },
    {
      "entropy": 0.06332132483366877,
      "epoch": 8.434405252030711,
      "grad_norm": 0.64453125,
      "learning_rate": 1.1691111111111113e-05,
      "loss": 0.0382,
      "mean_token_accuracy": 0.9891861483454705,
      "num_tokens": 126835959.0,
      "step": 4740
    },
    {
      "entropy": 0.0640255069010891,
      "epoch": 8.452208745966395,
      "grad_norm": 0.62109375,
      "learning_rate": 1.1668888888888889e-05,
      "loss": 0.0377,
      "mean_token_accuracy": 0.9889223039150238,
      "num_tokens": 127100029.0,
      "step": 4750
    },
    {
      "entropy": 0.06444245052989572,
      "epoch": 8.470012239902081,
      "grad_norm": 0.59375,
      "learning_rate": 1.1646666666666668e-05,
      "loss": 0.0379,
      "mean_token_accuracy": 0.9890415463596582,
      "num_tokens": 127364826.0,
      "step": 4760
    },
    {
      "entropy": 0.0633338522980921,
      "epoch": 8.487815733837765,
      "grad_norm": 0.53125,
      "learning_rate": 1.1624444444444446e-05,
      "loss": 0.0376,
      "mean_token_accuracy": 0.9891131032258272,
      "num_tokens": 127631132.0,
      "step": 4770
    },
    {
      "entropy": 0.06168914858717471,
      "epoch": 8.505619227773451,
      "grad_norm": 0.64453125,
      "learning_rate": 1.1602222222222223e-05,
      "loss": 0.0368,
      "mean_token_accuracy": 0.9892706379294396,
      "num_tokens": 127903917.0,
      "step": 4780
    },
    {
      "entropy": 0.06199188134633005,
      "epoch": 8.523422721709135,
      "grad_norm": 0.5625,
      "learning_rate": 1.1580000000000001e-05,
      "loss": 0.037,
      "mean_token_accuracy": 0.9891880311071872,
      "num_tokens": 128173872.0,
      "step": 4790
    },
    {
      "entropy": 0.06130934236571193,
      "epoch": 8.541226215644821,
      "grad_norm": 0.63671875,
      "learning_rate": 1.1557777777777779e-05,
      "loss": 0.0375,
      "mean_token_accuracy": 0.9892963591963053,
      "num_tokens": 128444160.0,
      "step": 4800
    },
    {
      "epoch": 8.541226215644821,
      "eval_chemistry_entropy": 0.20646600383520125,
      "eval_chemistry_loss": 1.2259459495544434,
      "eval_chemistry_mean_token_accuracy": 0.810901032447815,
      "eval_chemistry_num_tokens": 128444160.0,
      "eval_chemistry_runtime": 34.6316,
      "eval_chemistry_samples_per_second": 14.438,
      "eval_chemistry_steps_per_second": 3.609,
      "step": 4800
    },
    {
      "epoch": 8.541226215644821,
      "eval_physics_entropy": 0.2020608036518097,
      "eval_physics_loss": 1.2773762941360474,
      "eval_physics_mean_token_accuracy": 0.8050094060897827,
      "eval_physics_num_tokens": 128444160.0,
      "eval_physics_runtime": 41.3256,
      "eval_physics_samples_per_second": 12.099,
      "eval_physics_steps_per_second": 3.025,
      "step": 4800
    },
    {
      "entropy": 0.06136176576837897,
      "epoch": 8.559029709580505,
      "grad_norm": 0.5859375,
      "learning_rate": 1.1535555555555556e-05,
      "loss": 0.0362,
      "mean_token_accuracy": 0.9893068693578243,
      "num_tokens": 128722048.0,
      "step": 4810
    },
    {
      "entropy": 0.06203246309887618,
      "epoch": 8.57683320351619,
      "grad_norm": 0.63671875,
      "learning_rate": 1.1513333333333334e-05,
      "loss": 0.0371,
      "mean_token_accuracy": 0.9892611719667912,
      "num_tokens": 128992062.0,
      "step": 4820
    },
    {
      "entropy": 0.061620937939733264,
      "epoch": 8.594636697451875,
      "grad_norm": 0.59375,
      "learning_rate": 1.1491111111111113e-05,
      "loss": 0.0366,
      "mean_token_accuracy": 0.9894390199333429,
      "num_tokens": 129268293.0,
      "step": 4830
    },
    {
      "entropy": 0.06397192520089448,
      "epoch": 8.61244019138756,
      "grad_norm": 0.66015625,
      "learning_rate": 1.1468888888888889e-05,
      "loss": 0.0383,
      "mean_token_accuracy": 0.9887047711759805,
      "num_tokens": 129531930.0,
      "step": 4840
    },
    {
      "entropy": 0.06182295905891806,
      "epoch": 8.630243685323245,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1446666666666668e-05,
      "loss": 0.0368,
      "mean_token_accuracy": 0.9893895827233792,
      "num_tokens": 129802790.0,
      "step": 4850
    },
    {
      "entropy": 0.0626299069263041,
      "epoch": 8.64804717925893,
      "grad_norm": 0.87890625,
      "learning_rate": 1.1424444444444444e-05,
      "loss": 0.038,
      "mean_token_accuracy": 0.9892073828727007,
      "num_tokens": 130068311.0,
      "step": 4860
    },
    {
      "entropy": 0.061919541517272594,
      "epoch": 8.665850673194614,
      "grad_norm": 0.67578125,
      "learning_rate": 1.1402222222222224e-05,
      "loss": 0.0382,
      "mean_token_accuracy": 0.988821929320693,
      "num_tokens": 130331797.0,
      "step": 4870
    },
    {
      "entropy": 0.06368331348057836,
      "epoch": 8.6836541671303,
      "grad_norm": 0.625,
      "learning_rate": 1.138e-05,
      "loss": 0.0377,
      "mean_token_accuracy": 0.9892549213021994,
      "num_tokens": 130597714.0,
      "step": 4880
    },
    {
      "entropy": 0.06053618721198291,
      "epoch": 8.701457661065984,
      "grad_norm": 0.6328125,
      "learning_rate": 1.1357777777777779e-05,
      "loss": 0.0365,
      "mean_token_accuracy": 0.9893568214029074,
      "num_tokens": 130872668.0,
      "step": 4890
    },
    {
      "entropy": 0.06338407367002219,
      "epoch": 8.71926115500167,
      "grad_norm": 0.54296875,
      "learning_rate": 1.1335555555555558e-05,
      "loss": 0.0381,
      "mean_token_accuracy": 0.9889190468937159,
      "num_tokens": 131134601.0,
      "step": 4900
    },
    {
      "epoch": 8.71926115500167,
      "eval_chemistry_entropy": 0.2058943647146225,
      "eval_chemistry_loss": 1.2301411628723145,
      "eval_chemistry_mean_token_accuracy": 0.8113660650253296,
      "eval_chemistry_num_tokens": 131134601.0,
      "eval_chemistry_runtime": 34.6478,
      "eval_chemistry_samples_per_second": 14.431,
      "eval_chemistry_steps_per_second": 3.608,
      "step": 4900
    },
    {
      "epoch": 8.71926115500167,
      "eval_physics_entropy": 0.2012183396220207,
      "eval_physics_loss": 1.2815872430801392,
      "eval_physics_mean_token_accuracy": 0.8050037651062012,
      "eval_physics_num_tokens": 131134601.0,
      "eval_physics_runtime": 41.2738,
      "eval_physics_samples_per_second": 12.114,
      "eval_physics_steps_per_second": 3.029,
      "step": 4900
    },
    {
      "entropy": 0.0626102042850107,
      "epoch": 8.737064648937354,
      "grad_norm": 0.6953125,
      "learning_rate": 1.1313333333333334e-05,
      "loss": 0.038,
      "mean_token_accuracy": 0.9891807556152343,
      "num_tokens": 131399285.0,
      "step": 4910
    },
    {
      "entropy": 0.061753800325095656,
      "epoch": 8.75486814287304,
      "grad_norm": 0.55859375,
      "learning_rate": 1.1291111111111113e-05,
      "loss": 0.038,
      "mean_token_accuracy": 0.9891921758651734,
      "num_tokens": 131665227.0,
      "step": 4920
    },
    {
      "entropy": 0.06289804736152291,
      "epoch": 8.772671636808724,
      "grad_norm": 0.6875,
      "learning_rate": 1.126888888888889e-05,
      "loss": 0.0379,
      "mean_token_accuracy": 0.9889983184635639,
      "num_tokens": 131932557.0,
      "step": 4930
    },
    {
      "entropy": 0.06295099146664143,
      "epoch": 8.790475130744408,
      "grad_norm": 0.67578125,
      "learning_rate": 1.1246666666666669e-05,
      "loss": 0.0376,
      "mean_token_accuracy": 0.9889470681548118,
      "num_tokens": 132199444.0,
      "step": 4940
    },
    {
      "entropy": 0.06066715130582452,
      "epoch": 8.808278624680094,
      "grad_norm": 0.53515625,
      "learning_rate": 1.1224444444444444e-05,
      "loss": 0.0368,
      "mean_token_accuracy": 0.98926189430058,
      "num_tokens": 132473847.0,
      "step": 4950
    },
    {
      "entropy": 0.06325976536609232,
      "epoch": 8.826082118615778,
      "grad_norm": 0.54296875,
      "learning_rate": 1.1202222222222224e-05,
      "loss": 0.0381,
      "mean_token_accuracy": 0.9890444248914718,
      "num_tokens": 132736489.0,
      "step": 4960
    },
    {
      "entropy": 0.06339925451902673,
      "epoch": 8.843885612551464,
      "grad_norm": 0.59765625,
      "learning_rate": 1.1180000000000001e-05,
      "loss": 0.0384,
      "mean_token_accuracy": 0.9889429863542318,
      "num_tokens": 133001446.0,
      "step": 4970
    },
    {
      "entropy": 0.0610861549153924,
      "epoch": 8.861689106487148,
      "grad_norm": 0.5703125,
      "learning_rate": 1.1157777777777779e-05,
      "loss": 0.0372,
      "mean_token_accuracy": 0.9893666543066502,
      "num_tokens": 133272419.0,
      "step": 4980
    },
    {
      "entropy": 0.0625727279810235,
      "epoch": 8.879492600422832,
      "grad_norm": 0.6015625,
      "learning_rate": 1.1135555555555557e-05,
      "loss": 0.0381,
      "mean_token_accuracy": 0.989263066276908,
      "num_tokens": 133538093.0,
      "step": 4990
    },
    {
      "entropy": 0.062265278515405954,
      "epoch": 8.897296094358518,
      "grad_norm": 0.828125,
      "learning_rate": 1.1113333333333334e-05,
      "loss": 0.0376,
      "mean_token_accuracy": 0.9890133492648602,
      "num_tokens": 133807882.0,
      "step": 5000
    },
    {
      "epoch": 8.897296094358518,
      "eval_chemistry_entropy": 0.20466419225931168,
      "eval_chemistry_loss": 1.2306396961212158,
      "eval_chemistry_mean_token_accuracy": 0.8113405866622925,
      "eval_chemistry_num_tokens": 133807882.0,
      "eval_chemistry_runtime": 34.6138,
      "eval_chemistry_samples_per_second": 14.445,
      "eval_chemistry_steps_per_second": 3.611,
      "step": 5000
    },
    {
      "epoch": 8.897296094358518,
      "eval_physics_entropy": 0.2000389884710312,
      "eval_physics_loss": 1.283610224723816,
      "eval_physics_mean_token_accuracy": 0.8052523717880249,
      "eval_physics_num_tokens": 133807882.0,
      "eval_physics_runtime": 40.9588,
      "eval_physics_samples_per_second": 12.207,
      "eval_physics_steps_per_second": 3.052,
      "step": 5000
    }
  ],
  "logging_steps": 10,
  "max_steps": 10000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 18,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.0407917889475277e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}